• Why 60% of S&P stocks are in a bear market

• Stocks to look (and not look) at and why

• Where we go from here (and how to position yourself)

60% of S&P 500 stocks are 20% or more below their all-time highs, yet we are within 1% of all-time highs as an index. (Morningstar)

How’s this possible? The tech sector is carrying the index. They're carrying the 'E' in the index's P/E, and that’s why the market still looks rationally valued.

The biggest stocks are earning the most, and the stocks that aren’t earning at record levels are being re-rated lower. See: MCD, LULU, NKE, etc. There are stock-specific reasons why these are down, but many non-tech stocks are down because the AI trade is sucking the air out of the room.

This makes sense. If I can buy NVDA growing at 70%, buying back hundreds of billions of dollars’ worth of shares a year, why would I buy anything else? But mechanically, within your portfolio, it can create an uncomfortable amount of risk.

So today I want to look at the types of companies to buy, which to avoid, and where we might go in the broader market from here.

What to Buy and What to Avoid?

In recent months, stocks like McDonald’s and Nike have become talking points amongst market participants.

Currently, MCD is down over 30% from its all-time high, and at its lowest valuation since the COVID crash in 2020. Nike is down 78% since its 2021 peak, and if you bought all the way back in 2013 (I was 10 years old!) you would be down on that investment.

This is a good warning against single-stock investing, but isn’t my actual point. Nike is down so much due to upstarts (Hoka, On) and consumer spending shifting away from its brand, while MCD is down due to fears surrounding the consumer, the conflict in Iran, and inflation.

If I had to bet on which decline is more structural, my money's on Nike. I think with the advent of Shopify, combined with the effectiveness of Facebook ads, the moat for legacy brands like Nike is near zero. You can buy much cooler, more affordable athleisure clothing with ease nowadays. Additionally, Nike still trades at over 22x earnings, a 30% premium to McDonald’s!

If we look at MCD, you can’t replicate its scale and worldwide footprint without spending billions of dollars. If the Iran conflict were to end, which it must at some point, McDonald’s and other inflation-impacted companies seem poised for a re-rate back to their historical multiples. While the conflict and oil crisis could continue and weaken the consumer further, I think demand for food is a lot more inelastic than demand for athleisure.

I’m not calling for a bottom in McDonald’s here, just saying that if I'm shopping today, I'd focus on names that have struggled because of shorter-term issues.

While Nike could be ripe for a turnaround due to new management coming in, and new marketing efforts, I believe bottom fishing in stocks similar to McDonald’s would yield better forward returns from here. I think if we get MCD around a 15x forward P/E ratio (~$205/share), that might make it a no-brainer.

Where Do We Go From Here?

Before I wrap this article up, I want to discuss what happened in the 2000 dot-com bubble, specifically to the same cohort of S&P 500 stocks that were in bear markets near the top.

While it’s hard to disaggregate which stocks were specifically in bear markets in 2000-2002, what we can do is look at equal-weighted S&P returns vs. market-cap-weighted returns. This demonstrates the benefit of diversifying away from the top-performing stocks, and allocating equally to lesser-performing ones (think NKE and MCD today).

Cumulatively, from the 2000-2002 unwind, you outperformed the market-cap-weighted S&P by 27 percentage points if you allocated to the equal-weighted S&P. Instead of losing 37.6% of your money, you’d have lost only 10.6%. In a 37.6% drawdown, you need a 60% return to get back to even. In a 10.6% drawdown, you only need a 12% return to get back to even.

I’m not saying that today’s market is the same as the dot-com bubble. There are countless reasons why “this time is different.” OpenAI’s and Anthropic’s revenue charts are parabolic. We are on the path to artificial superintelligence. Every day I am astounded by what AI can do.

I only bring this topic up because if you’ve performed really well over the past few years, allocating a higher % of your portfolio to more conservative names could make sense. Obviously, it’s a little ironic that I’m writing this after posting an article about how I’m bullish on Oracle and its AI efforts, but I think a barbell approach might make sense here.

That said, I am still very much exposed to the AI trade within my own portfolios, but I’m open to increasing my more conservative exposures in an attempt to mitigate larger drawdowns.

This is not financial advice.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or checking out my most popular posts:

1. Why this opportunity exists

2. The improving economics of AI Capex & Datacenter Spend

3. Long Compute (via Oracle)

Why This Opportunity Exists

Oracle has recently become something of a battleground AI stock, for a few reasons:

1. A few months back, Oracle’s 5-year CDS went from roughly 145 bps to ~215 bps, a ~70 bp increase. A CDS is essentially insurance against Oracle defaulting on its debt. So if you own Oracle bonds, you can buy a CDS that pays out if Oracle defaults. When that insurance gets more expensive, the market is pricing greater credit risk.
   
   In simple terms, the market was saying that Oracle has become much riskier to lend to.

2. A ballooning Remaining Performance Obligations figure. Their RPO has grown 46% YoY, and Oracle is also claiming that they will turn 50% ($332b) of that into revenues over the next 3 years.
   

3. Concentration risk to OpenAI. ~45-50% of their RPO is contracted with OpenAI. People draw comparisons of this AI buildout to the dot-com bubble, and there is fear that having so much revenue tied to one customer could cause Oracle’s demise.

4. Force Majeure notice on Project Jupiter: Last week, Oracle sent a force majeure notice to the developer of Project Jupiter, its Stargate data center in New Mexico built to serve OpenAI. In plain terms, the notice lets Oracle push back lease payments if the site misses its 2028 target, which is looking likely given the power and permitting delays. The stock dropped 3% on the news. For me, this read more like a timing thing, not them walking away from the site. They are just covering their liabilities if there is a gap in payment timelines.

For these reasons, Oracle has become somewhat of a dog in the AI trade. Where other compute providers (CRWV, NBIS) have rallied, Oracle has stalled, and in fact it’s down ~55% from its 52wk high.

I honestly think ORCL is one of the simplest longs in the market. To like the stock I think it comes down to a few beliefs you have:

• Compute is valuable

• AI usage is only going to increase over time

• Concentration risk does not matter. Even if OpenAI were to go bankrupt (low probability), that compute can be sold to another AI lab (Anthropic, Google, Meta, etc.)

• Agentic AI unlocks a previously underrepresented customer for AI, the consumer. This demand could be another major driver of compute consumption over time, irrespective of which agentic product wins (Muse, Grok Bot, Instinct, Dots, etc.)

With those beliefs it’s easy to underwrite the bull case. What’s particularly interesting to me is that if you look at semis/proxies for the AI bet, the implied sentiment around the AI trade seems positive right now. All except for Oracle.

What this means is that a lot of players are offsides, and there are many incremental buyers of Oracle as these fears dissipate.

Right now I think the market is being a little myopic on Oracle. They see one major customer and are scared of concentration risk. I see strong demand, evidence that older hardware retains earnings power, and revenue that will far supersede the cost of capital on these infrastructure investments.

I believe the increase in WACC of these hyperscalers is more so a function of the sheer scale of how much capital they need to fund this buildout, not an indictment on the feasibility of these investments.

Capex Validation: Hyperscalers are Seeing Improved Economics on AI Spend

Before we move into valuation and how much upside I think there is in Oracle, I want to go over some quick data points that serve to validate Oracle and other hyperscalers’ capital expenditures.

1. GPU rental rates are increasing

2. GPU depreciation schedules are longer than the AI skeptics have assumed

3. Amazon is saying their payback period on a datacenter is 3 years!

All of these developments lead to an environment where the market should appreciate existing compute as well as current investments in expanding compute capacity.

I say should because for some companies it already has; see MSFT as an example. From the recent broader AI unwind in late July, MSFT is up over 30%.

I believe this trend will only continue, as the market begins to appreciate heightened capital expenditures, even at the cost of near-term FCFs.

Michael Mauboussin recently wrote a great piece on FCF, ROIC vs. WACC, as well as ROIIC (Return on Incremental Invested Capital), which adds more color to this thesis.

One of my main takeaways was that on a relative basis, Oracle will have the largest improvement on their incremental invested capital.

On longer time horizons, returns on invested capital ≈ returns on stocks. With this in mind, we should ideally buy stocks in companies that are making positive capital investments.

While near-term FCF looks dire, taking a longer lens we can see that their ROIC profile remains strong, and that these major investments seem likely to pay off given the developments in GPU prices that I highlighted earlier.

Michael Mauboussin highlights a good example of how burning FCF in the name of investments can make for exceptional returns with Walmart:

On top of all of this, Larry Ellison is closely tied to the Trump admin, and if we were to see any analogues to the “Too Big To Fail” in this AI investment cycle, it seems likely Ellison could be one of the beneficiaries.

Additionally, if OAI concentration risk is the primary Oracle bear case, is their recent acceleration to a $70B ARR not dispelling that?

Investment Thesis

The investment case here is pretty simple: Oracle turns its contracted demand into rapidly growing revenue and earnings, and the stock re-rates as they deliver on that story.

Oracle’s long-term financial outlook targets $130 billion of revenue in FY2028, $185 billion in FY2029, and $225 billion in FY2030, alongside non-GAAP EPS of $10.65, $16, and $21, respectively. These are management targets, but they show you the scale of the earnings opportunity if Oracle executes.

There is already evidence of that conversion. In Q1 FY2027, infrastructure revenue grew 121% to $7.4 billion, supported by 850 MW of additional delivered capacity. Oracle also raised its FY2027 revenue outlook to at least $90 billion. The story is already progressing from signed contracts into operating revenue.

Bull, Base, Bear Cases

If we assume for a second that management hits on these targets, and FY2030’s EPS is ~$21, and the stock trades at 20x earnings, we get a stock price of $420 in mid-2030. At a 10% discount rate, that backs out to a fair value of $295 today, so over 2x upside from today’s levels.

If we lower our assumptions, say to 2030 EPS of $15, so 70% of their targets, and an 18x multiple, we get $270 or roughly $190 discounted back to today.

If we cut management’s figures in half, and say they only get $10 of EPS by 2030, assign a 15x multiple to that, we get $150 which is worth about $105 today.

So baking these all in, we have a ~2x on the bull case, +38% on the base case, and -24% on the bear case. That seems like a palatable risk profile even if you assign a high likelihood to the bear case, which I do not.

To be clear, I don't think it's a foregone conclusion that they perfectly hit $21 of EPS in fiscal year 2030. I used that as a framework for backing out to today's prices and building an investment case. I think there are many scenarios where they far exceed that $21 of EPS and others in which they do not. Additionally, in 2030, if they hit on $21 of EPS it’s entirely possible a 20x earnings multiple is too conservative. It all comes down to your beliefs on the validity of these investments in AI infrastructure.

The Trade

I am invested in ORCL via common stock as well as longer dated call options. I have December 2027 calls struck at $230/share to play a more aggressive re-rating, but I also have LEAPS for December 2028 struck at $160/share.

The $230s are a bet that the re-rate comes much faster than I expect, and I sized it much smaller because I view it as much less likely. The common stock and $160 LEAPS are the true reflection of my investment case.

I do not foresee any near-term catalysts that will change the market's view. I believe they must execute and deliver on the buildout, and the market will appreciate it when the revenues and growth come.

Disclaimer:

I hold investments in Oracle

This is not financial advice, do your own due diligence before making any investment decisions.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or checking out my most popular posts:

In this article I will:

• Give you a better framework of understanding why a stock is up or down

• Help you stack bets in your favor (Meta as an example)

Stocks as a Bundle of Bets

This graphic visualizes what comprises a stock and its risks. These subsections within a stock are not static in size, as sometimes the market can hone in on one factor or risk and that can drive the stock for a period of time. (see SaaS vs. Claude earlier this year)

Different stocks also contain different levels of each bet, as you can see a deep value cigar butt stock trades on its value, where an AI infrastructure company moves primarily with its sector. Over longer time frames, the value subsection typically grows in size, as the old Benjamin Graham saying goes:

“In the short run, the market is a voting machine, but in the long run, it is a weighing machine.”

In shorter time frames, most of the move comes down to how the market is moving, not really stock specific bets.

This phenomenon is why I prefer to invest over longer time frames. When so much of the market is moved by hard to predict factors (geopolitics or interest rates) it’s easier to take a longer lens and ask, “Will this company be around in 5 years? Will they sell more or less of their products? Will their investments pay off?”

Stacking Bets in Your Favor

Coming back to the original image, I think it’s important to consider all of these bets when entering a position:

For example, I have been really bullish on Meta for the past nine months. In that time, it sold off to the low $500s, and it became a hated stock, as it commonly does. I had a positive view on the business, its products, and growth prospects going forward. I also had a positive view on the sector as a whole, believing that hyperscalers would lead to the next leg up in the AI trade. (I still believe this)

All of this leads to me having a positive view on the business (red), sector (green), and market (blue) sections of the stock bet bundle.

In addition to that, the public sentiment was near all time lows. Mark Zuckerberg was getting clowned on like he had been for the Metaverse capital expenditures. This meant that there was no momentum in the price, and that most players were not long the stock.

It also meant that the yellow and purple sections (how players are positioned) worked in my favor, because when a stock is hated, it means many people don’t own it, and there are a lot of incremental buyers as positive news develops.

That’s what drove a rapid re-rate for Meta over the past month, as it’s currently up over 37% in that period.

Because Meta was so hated, when the sentiment shifts, as it has recently, the recovery is quite violent. This is similar to how Google was “behind in AI”, and “losing share in search because of ChatGPT” in 2025. We all saw how that played out.

I don’t write this to brag about being long Meta in this time. It was an uncomfortable hold, and there were times when I doubted myself. I am still long Meta, but I recognize that now the bar for their execution has been raised as the stock has become more in favor with market participants.

I write this to highlight these two situations as examples of how betting against the consensus can be so rewarding.

Knowing Why Your Stock is Down

Back to my brother’s question. Most of the time, I still don’t know exactly why a stock is down. And I don’t think it’s that important to actually know, unless it has to do with your thesis.

If the broader market is down on geopolitics or rates, that doesn’t affect my longer thesis on Meta. If anything it can be an opportunity to buy more.

Now if Meta specifically is down on a negative headline, that would be something to dive deeper into, and assess whether it has any effects on my thesis.

All of this is to say, when you enter a position, try to stack as many bets as you can. Find a good stock, in a strong sector, and it’s even better if it’s a hated stock.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or checking out my most popular posts:

• What I got right and wrong about Meta

• How they are executing now (hint: very well)

• Where I see them going in the future

What I got right and wrong so far:

Back in July, as part of an exercise I got from a great Michael Mauboussin article, I made a few predictions in a Google Doc about Meta. Here they are:

Instead of October/November, Muse Spark 1.1 released just three days later and was near/at the frontier.

Instead of Q1 2027, Muse, Meta’s personal agent for consumers, launched on September 8th.

The third prediction has come to pass, but not how I envisioned it. They have opted for more of a usage-based model, serving Muse Spark 1.3 on platforms like Cursor and Muse Code. While I haven’t tried Muse Code specifically, I would like to see more seamlessness across Meta AI/Muse (the personal agent)/Muse Code. Right now it seems cluttered and messy.

Just look at this interface compared with the likes of Claude Code & Codex. Those are much more user-friendly, straightforward and seamless with the base AI subscriptions you get with Claude and ChatGPT. As a completely non-technical person, I see some room for optimization & continuity across their AI products.

Additionally, why am I unable to pick between Muse Spark 1.1 and Muse Spark 1.3 on the standalone Meta AI app? Just seems like there is room for a Meta AI/Muse Code app merger of sorts.

These criticisms aside, I do want to give praise where it’s due, and that’s with Muse.

How Meta is Executing Now: Muse

I have been excited about the potential of a consumer-focused OpenClaw variant for some time now. Back in Q1 of 2026, everyone was buying Mac Minis to have an agent live on, where you could text it to help you book reservations or trips.

Now, Meta offers it as a free app. Aside from the price point of $0, it’s also a seamless user experience. I am very AI-forward, yet I would never take the time to set up a Mac Mini and have it host an agent for me. My mom, who has never used ChatGPT or Claude, was super excited to hear about an agent that could help book flights or hotels for cheaper.

Now I can download a free app to do what people were buying Mac Minis for just six months ago.

In short, I think OpenAI and Anthropic, while racing to win the enterprise, left the consumer behind. In their place, startups like Instinct and Poke demonstrated that there was demand for this sort of product. But with Muse, Meta has both the scale advantage & the model to deliver personal superintelligence and give you a virtual machine the agent can use to browse the web.

This allows you to do really cool things that are harder to do with ChatGPT and Claude. Or you can, but it burns through all your usage limits. Some use cases I have seen/enjoyed so far:

• Texting Muse and having it schedule a DMV appointment for me

• A periodic flight tracker that tracks prices for all my favorite destinations

• A Twitter feed optimizer, finding the tweets it thinks I will care about the most, sorting through the engagement bait

• Pay traffic citations

• Finding underutilized subscriptions

• Creating podcasts on interesting subjects

• Calling AT&T to lower your bill (I don’t have access to this functionality yet but it will be a game changer once I get it)

There are countless other ways to use agentic AI, but overall giving away tons of free tokens plus access to a virtual machine allowed Muse to provide use cases that are significantly differentiated from current LLMs.

There are some interesting takeaways from this, namely the impending doom for businesses that rely on human laziness and complex cancellation flows. Your agent doesn’t get tired or lazy. If you tell it to cancel XYZ subscription, it just will.

Meta has really hit on something with this consumer-focused project, and while the path to monetization is not as clear as just selling tokens for $20/$100/$200/month, I think they can figure it out.

In a recent podcast, Zuckerberg specifically mentioned that there is potential to implement something of a take-rate business model where Meta could take a small percentage of money saved/made by Muse.

In addition to all of these positives, Meta has the benefit of having one of the best distribution vehicles possible. Instead of showing ads, they can advertise Muse to their 3.6 billion MAUs.

This distribution advantage is already showing up in the results. Just look at the download figures for Muse so far:

It’s also important to highlight their LLMs’ competitiveness in terms of pricing with other frontier models.

It’s even clearer to see this when we look at usage on OpenRouter, where Meta’s Muse Spark 1.3 has taken 43% share!

I am very curious to see how this continues, especially if Meta’s next model, Watermelon, has a major intelligence lead over GPT-6 Sol or Anthropic’s upcoming models, because we have seen Meta is willing and able to sell their tokens closer to their cost, significantly undercutting OpenAI and Anthropic for similar (or even higher) levels of intelligence.

This could be part of a strategy to hit OpenAI's and Anthropic's ARRs ahead of their upcoming IPOs and suck the air out of the room for further valuation rounds. This is a sound strategy considering we have seen a clear cycle emerge: more compute = stronger models = increased market share & revenues = easier equity raises = more $ for compute.

This has occurred at the same time that Meta has slowed their previous “tokenmaxxing” agendas and I would not be surprised to big customers’ spending being listed as a ‘Business Risk’ in Anthropic’s upcoming S-1 filing. A brief synopsis of the tokenmaxxing saga below:

What’s on the horizon?

This is all good to hear, and the pace at which Meta is iterating is very promising. But what’s more is that they are also not succumbing to the fear-mongering being pushed by the OpenAI/Anthropic cohorts. I thought this tweet from Zuck was pretty aggressive and I liked his rhetoric in response to the “pacing of the frontier” that Sam and Dario have recently called for.

The standout quote:

“Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built.”

On the horizon, Meta is close to releasing Watermelon, and they've already begun work on the model after it. In addition to that, they have already begun work on the model after Watermelon.

What I hope to see from Meta is continued iteration on their AI product suite, and hopefully that can lead to re-rating upwards, as it feels like much of their underperformance in 2026 has been due to the market’s mistrust of their current capex policies.

Additionally, their Meta Connect event is just next week, and I expect that they will have even more positive developments there. I think there is a clear opportunity to have their models live on their wearables, and provide functionality where you can interface with Muse through voice.

If Muse adoption continues to be strong, and they can further monetize their investments in AI, I believe a sharp re-rate back to premium big tech multiples of 25-30x is well within the realm of possibility.

That, coupled with strong growth estimates for their core ad business is why I continue to view Meta as one of the best opportunities in the stock market right now.

It does finally seem like some of Meta’s AI efforts are being appreciated, with the stock up ~25% since August 18th. My hope is that this is the beginning of a bigger move, as we have seen some previous moves like this fade over time.

Disclaimer:

I do hold investments in Meta Platforms.

This is not financial advice, do your own due diligence before making any investment decisions.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or checking out my most popular posts:

Before we get into the details, here’s my favorite visual to demonstrate the power of small gains (or losses) over time.

Improving 1% each day for a year leads to ~3700% gain. Conversely, 1% decline each day over the course of a year leads to a 97% loss. Let’s find those 1% improvements.

Expense Ratio Gaps

This one is very simple, but if you take a quick look into what you hold, it can save you thousands of dollars over the course of your investment lifetime. As an example, let’s look at SPY vs. VOO, two S&P 500 ETFs.

As you can see from this table, investing in SPY, which tracks the same index as VOO, costs you. At shorter time horizons, the gap is minimal. But as that annual fee gap compounds, the effects do as well.

This effect can cost you well over $100k if your starting balance is higher as well, and this table is understated as it assumes no continued investments are made each year. If we assume that we invest $10k a year, look what happens.

This doesn’t just apply to broader market ETFs, and this is a scenario where AI tools (ChatGPT & Claude) can be super helpful. If you want exposure to gold, for example, you might search and find GLD as the first ticker.

If you were to choose GLD over GLDM, you’re paying 4x the expense ratio for the same underlying gold exposure. And if you want to get even deeper into the weeds, GLDM has slightly outperformed GLD over a one-year period, delivering 32.93% returns vs. 32.54% for GLD.

So anytime you consider investing in an ETF, just take a look at the expense ratios of other ETFs to see if you’re getting the best deal for similar exposure profiles. It may not make a major difference over a short time horizon, but over time it adds up. You can ask ChatGPT or Claude for similar ways to play a trade that cost less in annual fees, but express the same idea.

Tax Lots and Minimizing Your Tax Bill

Let’s imagine you’ve bought a stock at three different times, and at three different prices.

• 3 years ago, you bought 100 shares at $50 per share (Lot A)

• 2 years ago, you bought 100 shares at $120 per share (Lot B)

• 6 months ago, you bought 100 shares at $130 per share (Lot C)

With the stock now trading at $150, selling any of these 100 share lots brings in $15,000. But, depending on which lot you sell, you are realizing different levels and types of capital gains. Here’s how selling each lot would differ in federal tax amounts:

Selling Lot B gives you the same $15,000 in cash today, but produces $1,050 less in taxes for you to pay. Most times, it pays to wait until you’ve held your investment for over a year and select the highest purchase price lot to minimize capital gains tax.

I think most people are familiar with tax loss harvesting, but the basic idea is that you sell an investment at a loss, and use that loss to offset some amount of capital gains you would otherwise need to pay. As an example:

Selling Options to Get Into and Out of an Investment

I’ve saved this for last because I view it as my favorite method listed here, but it’s also the most complicated & controversial. I will attempt to make this as simple as possible but it’s important that I state: options are much more complicated than stocks and you should fully understand how they work before using them in your portfolio.

I think the only way to explain this method is through an example, so here goes.

Imagine you want to buy Robinhood ($HOOD) at $115/share. Let’s say you have strong conviction it’s a great company and you are willing to buy and hold it, with $115/share being your cost basis.

To enter into a position in $HOOD, you could either:

• Buy stock

• Buy a call option

• Sell a cash-secured put

Selling a cash-secured put means agreeing to buy 100 shares of Robinhood at $115 per share if assigned, while setting aside enough cash to make that purchase. Assignment can happen before or at expiration.

In return for signing this contract, you receive a premium, aka cash today. Let’s take a look at the options chain to illustrate this more clearly.

In this example, you’re agreeing to buy 100 shares of Robinhood at $115 per share if assigned on the put, which expires April 16, 2027.

So why do this over just buying stock?

If you sell this option, you receive $2,135 upfront and reserve enough cash to buy the shares if assigned. If Robinhood finishes below $115 at expiration, you should generally expect assignment, although it can happen earlier. The premium makes your effective purchase cost $93.65 per share before fees.

If you’re set on owning Robinhood anyway and would still want to hold it at $93, assignment gives you the shares you wanted. You also keep the $2,135 premium you received when you sold the put.

But what if Robinhood rallies and you’re never assigned? You keep the premium, but you don’t acquire the shares and could miss gains larger than the premium you collected. That’s one risk of selling a cash-secured put to get into a stock.

If you’re assigned, you keep the $2,135 premium you received and own the shares that you wanted to anyway. It helps lower your cost basis. If you can lower your cost basis, you can enhance your long-term returns.

Suppose Robinhood does really well over the next 5 years, delivering 20% annual returns from $115/share to about $286/share. That’s a great return on the shares. But, if you got assigned on that put, your effective cost basis would be $93.65. Consequently, your return wouldn’t be just 20%/year; it would be 25%/year. A pretty substantial step up.

But that’s not it, because you can enhance your returns on the other end as well. If you want to sell your shares you have two options:

• Sell your shares normally

• Sell a covered call

Selling a covered call means you’re agreeing to sell your shares at a predetermined price by a certain date in the future. It’s very similar to a put, except the risk manifests itself differently.

In return for selling a covered call, you receive a premium and your upside is capped. Let’s use Google as an example.

If you sold this call on Google, you’re agreeing to sell your 100 shares at $330/share by April 16, 2027. Your premium in return is $39.60/share or $3,960 per 100 shares.

This means that you start missing out on gains above $369.93 per share if Google goes above that before April 16, 2027.

This is why I mentioned these methods of selling options are controversial. The risks manifest themselves in unconventional ways. However, in my view, you have the discretion to take these risks, and if you have strong conviction in a stock, selling a cash-secured put makes sense. Similarly, if you are happy exiting a stock at a certain price, covered calls make sense to get out of a stock.

Coming back to the HOOD example, I’ll go through what the total return picture might look like if you sold a covered call to exit the position.

So we need to figure out how much we’d get paid to sell a covered call with HOOD trading at $286/share. If we assume we sold a similar call to the Google example I just showed, the ratio of premium/share price was 12%. You get $3,960 to sell Google at $330/share. Applying that same ratio (a massive oversimplification, I know) you would get $3,432 to sell a covered call at $286/share in HOOD.

This is way oversimplifying things, but I want to show you how much (roughly) you can enhance the annualized returns by selling options on the way in and out of a position.

So, earlier we went from a 20% annualized return to 25% annualized by selling a cash-secured put to get into the position. If we sell a covered call, that goes from 25% annualized to ~28%. In total, we get ~8% of difference per year by selling the options on the way in and out, with the stock’s return staying the same.

20% vs. 28% annualized is a big difference. A 20% annualized return doubles every 3.8 years. 28% doubles every 2.8.

Now the issue is this makes a lot of assumptions:

1. It assumes the stock will go below your cash-secured put’s strike price, and then go back up eventually (if the stock continues to go down, you’re still better off having sold the put rather than buying the shares outright)

2. It assumes the stock will go past your covered call strike, if it doesn’t you will still hold the shares (not a terrible outcome)

3. It assumes you have enough capital and discipline to sell these options and keep the position open when the unrealized profit and loss may be in the red.

Actionable Takeaways

For most readers, the first two sections are the easiest to put into practice. The last section, and the practice of selling options involves a level of complexity that is probably too much for most people who just want to passively grow their wealth over time.

To get immediate value from this article, I suggest uploading your portfolio to Claude or ChatGPT, and asking, “For each of my holdings, is there a cheaper fund with the same exposure, and how much would switching save me over 20 years?”

Additionally, if you’re interested in a bit more color on Robinhood, I wrote it up on Tuesday, 9/8/26, here:

Lastly, if you have the time & energy for it, I am a proponent of selling options, but you should know it will require more time from you. There are also unconventional risks that present themselves when using options. If you do decide to use them, I would recommend doing much more research.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or checking out my most popular posts:

Now at ~$120, and a $110 billion valuation, I have to determine if that price is worth paying. I am a believer in what they are doing, offering higher interest rates than peers on uninvested brokerage cash, targeting younger customers, and growing a substantial subscription base for Robinhood Gold. On top of that, they offer a great credit card, are rolling out access to bank accounts, and plan to offer mortgages eventually.

The Business

To me, Robinhood is a very compelling fintech company. Apart from the 2021 GME crisis, I think they have done a great job, and their product lineup covers most of what a retail investor wants. They’ve become a mostly one stop shop for taxable, retirement, and managed accounts. In addition to all of these accounts, they make passive investing frictionless, and their Robinhood Gold is a steal at $5/month. They are even growing a social media on the side. (You can follow me at NPV on Robinhood Social.)

These highlights fail to mention what they already do in crypto, prediction markets, and whatever tokenized stocks are used for. I mentioned these last because while they contribute to their revenue (for prediction markets, pretty substantially,) I do not rate them as part of my longer term investment thesis for a few reasons:

1. Crypto is wildly cyclical, incredibly volatile, and I’m just not a major believer long term.

2. Prediction markets seem ripe for regulation in upcoming election cycles.

3. I haven’t seen a compelling enough use case for tokenized stocks to assign them meaningful value in my thesis.

Revenue Breakdown

Notable highlights from this include:

• Event contracts revenue up 10x YoY

• Equities revenue up 95% YoY

• Robinhood Gold specific rev (the most sticky in theory) works out to $54M

• Gold subscribers are up 39% YoY

Options and equities generated roughly 36% of revenue this quarter. Add net interest income and Gold subscriptions, and that reaches about 70%. These are the businesses I’m more comfortable building a long-term thesis around, although I wouldn’t assume this quarter’s earnings power holds through every market environment.

Robinhood’s Growth Engines

My investment thesis is that Robinhood can retain customers as their incomes and portfolios grow, and convince them to bring more of their financial lives onto the platform. There’s already clear evidence of this, with funded customers growing 7% YoY and Total Platform Assets growing 32% to ~$369 billion. Over the past 12 months, Robinhood brought in $75.7 billion of net deposits, equivalent to 27% of platform assets at the beginning of that period.

I view Robinhood Gold as another growth vehicle for them, as approximately only 17% of funded customers are Gold subscribers. That compares with 40% of new funded customers signing up for Robinhood Gold. That suggests to me that there is room for adoption to increase, especially as to fully pay for your Gold subscription at current interest rates, you need just under $2,000 in brokerage cash.

I could easily see Gold becoming an Amazon Prime-esque membership, where it’s a subscription that is far more sticky than others.

More assets are great, but they don’t automatically convert to proportionally more revenue. All else equal, Robinhood aims to attract customers who have more assets, but also who trade more. Higher trading activity in options, equities, and prediction markets is what Robinhood really wants. It seems like they’ll get it too.

Gen Z and Millennials, Robinhood’s primary target demographic, seem much more prone to gambling and more speculative investing.

Additionally, as older people age, and pass down wealth to younger generations I would posit that a much greater % of older Gen Z/Millennials have Robinhood accounts vs. these conventional investment platforms.

With all of these effects, if Robinhood retains these customers as their wealth grows, I expect substantially more assets on the platform over time, as well as trading activity.

Risks

Before getting into the investment case, it’s important to highlight that much of Robinhood’s revenues are interest rate sensitive. Net Interest Income makes up 30% of revenues and is the primary selling point for Robinhood Gold subs. If the Fed cuts, Robinhood would take a hit across multiple revenue segments.

What Would Justify Today’s Price?

As you can probably tell, I imagine Robinhood becoming a much larger business over the next decade. What I’m struggling with is how much of that future I’m already being asked to pay for. That becomes harder when some of its fastest-growing businesses are ones I’m reluctant to underwrite.

At present, they sit at a 42.4x forward price-to-earnings ratio, ~20% above their 3 year average PE ratio.

Instead of trying to guess at what each segment’s revenue will do over the next few years (you’ll find much better estimates of that from Wall Street analysts) I want to instead assume a future multiple, say in 2029 and work backwards to what their earnings would have to do to be a good investment from here.

Let’s start with the return threshold. For me, if we assume that investing in the broader market should give us ~10% annual returns on a nominal basis, I would like ~5% of additional annual return, giving me a 15% hurdle rate.

Now, let’s look at valuation ranges. I think it’s worth looking at 3 possible cases for this.

1. Robinhood has a 20x multiple in 2029. At half of the current multiple, it could bake in that the parabolic growth is done, and that the company is nearing maturity.

2. Robinhood has a 30x multiple in 2029. While it’s down from 2026, the company still has a positive sentiment in the market, and people think it will continue to do well.

3. Robinhood has a 40x multiple in 2029. Market sentiment is still overwhelmingly positive, and future prospects are strong.

At a 30x ending multiple, Robinhood would need to grow EPS by roughly 29% annually over three years to deliver my required 15% annual return. At 20x, the required EPS growth rises to nearly 48% annually. Even at 40x, EPS would need to grow around 17% per year.

If net profit margins and diluted share count stayed unchanged, revenue would need to grow at those same rates. Higher margins would reduce the revenue growth required, while additional share dilution would raise it.

Holding margins constant is definitely a simplifying assumption. Robinhood has already shown that it can grow revenue faster than expenses, as in 2025, revenue rose 52% while operating expenses increased 25%. If that operating leverage continues and translates into higher net margins, it could meet these EPS hurdles with less revenue growth than my table implies.

This makes it easier to assess Robinhood as an investment: am I comfortable with these growth assumptions?

To me, nearly 48% annual EPS growth, even with benefits from operating leverage, seems aggressive. The 29% and 17% hurdles seem more achievable, especially if margins continue to improve. But I still need to be comfortable with the combination of revenue growth and profitability required to get there.

Now, if we rewind the clock and look at what your assumptions need to be at a purchase price of $90/share, you can see the assumptions are much easier to underwrite.

We go from needing 29% annual EPS growth over the next 3 years to just 17% to get the same return on investment at a 2029 valuation of 30x 2030 EPS.

Takeaways & Action Plan

This exercise and its findings are why I stress the importance of entering a stock at a good valuation. The higher multiple a stock has, the more optimistic you have to be about its future prospects.

For me, to buy Robinhood at $120+ I’d need a clear-eyed view on what their long term margin profile will look like, and if the growth + margin expansion can support that EPS path I laid out.

For this reason I think the takeaway here for me is to add Robinhood to my watchlist, and keep an eye out for if it drops to ~$100 or below. However, if you have the brokerage cash available, selling a cash secured put might make sense. Looking at the April 2027 options chain, you can get ~$2400 in premiums currently selling one $120 strike put.

This puts your breakeven, or effective cost of owning the shares if assigned, at $96. Much closer to those $90 assumptions I laid out just a bit ago. Obviously, in this scenario you have to tie up $12,000 in capital, but if you’re a believer in Robinhood, this could make sense. If the put expires unexercised, you get your $12k back, and make ~20% over 220 days. The tradeoff here is that if something happens in the short term which materially hurts Robinhood, you’re going to be assigned at a higher price.

If you opt to wait, like I am, you can reassess before buying, but selling the put commits you to the trade. Additionally if Robinhood continues to run, you don’t get to participate in the upside. As always, this should not be treated as financial advice, and you should do your own due diligence. Options are inherently risky, and you should do more research than just reading this article before entering a position.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or check out my most popular posts:

Escape from Tarkov is a PvPvE extraction shooter. For the non-gaming initiated, this means that you bring in gear, attempt to find loot and valuables, and fight against players (real humans) and bots. After shooting and looting, your goal is to extract (survive) the raid and bring your spoils of victory back to your hideout.

After putting more hours into Tarkov than I’d like to admit, I’ve realized that I’ve learned many valuable lessons applicable to life and, more importantly for this Substack, finance.

As a result, and from seeing absurdist articles/videos doing well across social media, I thought I could share some of those lessons and relate them to the real world.

Lessons Learned: Escape From Tarkov

1. Bringing better gear into the raid increases your odds of success. Whether it be a better gun, or a helmet that stops more bullets, there is a reason to bring in better gear, even if it’s expensive.
   
   Relating this to real life, there’s a cost to preparing for something, whether it's a job interview, a presentation, or a sports game. In my view that price is worth paying in almost all cases.

2. Building on the last point, you can bring the best gear into raid, or do the most preparation for a job interview, and still fail. There is an asymptote to increasing your odds of success, and past a certain point you have to leave things to chance.

   

3. Sometimes things are outside of your control. In Tarkov, maybe that means you died to a cheater. In real life, maybe that manifests itself differently. You lose out on a job opportunity because it’s too far away, or someone better connected gets that chance, even if they’re less qualified.

4. In video games, as well as life, you cannot make progress without first taking risk. In Tarkov, there’s a hideout where you can craft things and buy and sell items on an in-game marketplace. You can make small amounts of money here from the safety of your hideout, but you can’t make substantial progress. To do that, you must go into raid, and fight other players and AI.
   
   Life is similar. You might be able to study something at home, learning what people have done in the past. You may even be able to apply some of what you learned. But to make real, measured progress, you need to act in the real world.

5. Sometimes it’s best to cut your losses. Oftentimes in a game like Tarkov, you can get pretty wounded, but still be alive. As a result, you’re forced to make a decision, continue on in your weakened state, or just return to your hideout and come back another day.
   
   Life, and finance specifically, is pretty similar. In real estate development, one of my professors told me of cases in which something along the development process went poorly, and as a result, they just handed the keys back to the bank and walked away, citing that “it’s not worth it to put good money after bad.”

6. More cooks in the kitchen doesn’t always lead to better outcomes. In Tarkov, there’s no team UI, which just means that you have to keep track of who’s who, and know where your teammates are to ensure you don’t end up fighting each other. As a result, there are some benefits to playing alone, namely that when you hear a footstep or gunshot that isn’t yours, you know it’s an enemy.
   
   In life, this can work similarly. Not in the sense that you don't know who's on your side. That doesn't really happen. But sometimes you get more done working alone.

   

7. Decisions fueled by frustration rarely work out. I had a friend who I’d commonly play Tarkov with, and sometimes he’d shoot at an enemy and miss. For most players, after missing a shot, they would just move on and continue about the raid as normal, as giving away your position makes it harder to fight that enemy.
   
   But for my friend, he would get annoyed, and try to fight them anyways. This commonly led to him pushing into an enemy who knew he was coming, where he was coming from, and how to set up and ambush him as a result of that.
   
   This lesson draws parallels to a quote from earlier: “don’t put good money after bad.” It’s a simple, yet effective rule for making better decisions. If you mess something up, sometimes it’s best to just walk away.

8. Mitigate risk whenever possible. In Tarkov, you can loot bodies of enemies that you kill, but it puts you at risk, as you are a stationary target. To combat this, you can crouch or lie down to make yourself less noticeable. You can also skip looting entirely to not take this risk.
   
   In finance, you can minimize risks similarly. By sizing positions small, setting stop losses, or using a margin of safety, you can manage your risk.

9. If your gut is telling you something, you should listen. After countless hours and raids in Tarkov, my gut was pretty well calibrated to tell me when and where I should expect to see enemies, or where I was most at risk of being shot at.
   
   Sometimes, I couldn’t explain exactly why I felt something was off, yet seconds later I was being shot at. What I’ve learned from this is to trust my gut. While this doesn’t apply one-to-one with finance, where feedback can be muddied, I think the takeaway is clear: trust that feeling.

10. There’s no way to guarantee success. Put differently, there is always risk of failure. With a game like Escape from Tarkov, you have many different ways of increasing your odds of success, but there are never guarantees. You can wear level six armor, an Altyn helmet (seen below), but there is no guarantee that these things will help you.

    

    Realizing and accepting that sometimes you will lose is a freeing experience. It’s much easier to move past a loss when you have the mentality that losing is part of playing the game in the first place. It’s what makes winning that much better.

11. People pay for novelty. In Tarkov, there’s a knife you unlock as part of prestiging (getting to a certain level and resetting your account). This knife has no utility, other than looking somewhat cool. As a result, I was able to sell it on the in-game market for a sum that vastly outweighed its utility.

12. Exclusivity increases value. There are in-game cosmetics (clothing) that require you to kill a boss 100 times. This grind is cumbersome and time-consuming, and it drives people crazy. As a result, people sell accounts for real life money that have this boss outfit.
    

    

    This is similar to a Veblen good, where demand goes up as price goes up, except the currency here is time instead of money. Nobody wants the outfit because it looks good. (In fact, it’s kinda bland compared to the newer clothing they have released.)

    They want it because it proves you killed the same boss a hundred times, and there's no way to fake that. If this thing took 10 kills to unlock, you'd see it everywhere and nobody would pay for an account to get it.

All of this is to say, a few things stood out once I had them all in one place:

• Take action, and more importantly take risks, because that’s how you can make real progress

• Become comfortable with loss, and the fact that not everything is in your control

• Come prepared, but accept that there are limits to how much you can increase your odds of success

• Don’t put good money after bad

I plan to turn this into somewhat of a series, so if you enjoyed this, and/or want to hear my takeaways from games like Rust, Settlers of Catan, etc., make sure to subscribe.

Subscribe now

Enjoyed this article? Consider subscribing, leaving a comment, or check out my most popular posts:

A Buffett-style value manager would argue diversification is protection against ignorance, and that’s right to a certain extent, but good businesses can suffer undeserved drawdowns.

I already know my own beliefs based on my portfolio, but I think going through and arguing for and against both sides can be a useful exercise for anyone unsure of how to tackle this in their own portfolio.

A Coin Flip: Martin Shkreli on Kelly

The catalyst for me discussing this topic was a conversation that Martin Shkreli had on one of his daily livestreams. I don’t frequent his livestreams, but saw a clip on YouTube and it piqued my interest.

Shkreli, for the past year or so, has been building a Bloomberg competitor, dubbed Godel Terminal, and he recently made a Kelly Criterion widget that serves to demonstrate the Kelly Criterion in a more visual format.

The Kelly Criterion is a mathematical formula used to find the optimal bet sizing for a series of bets/investments to grow wealth as fast as possible.

I won’t go too into the weeds of the Kelly Criterion and the tool, as the purpose of this article is about applying these frameworks to portfolio management, but the main takeaway is that you should be betting much more conservatively than you would think.

It’s also important to understand that stocks are not perfectly analogous to the coinflip example that Shkreli utilizes here for a few reasons:

1. Payoff profile is not the same, betting on a coin flip you risk 1 unit to gain 1. Your payoff is 1:1 but you lose everything you bet when you lose.
   
   In equities, maybe the downside is 30%, 50%, 70%, it’s really dependent on stop loss, valuation floor, the list goes on. The upside is similar. For a multi-year position, the upside could be 2x, 5x, 10x, or even more. For a short trade, maybe it’s 5 or 10%.

2. Also Shkreli’s coin flip game assumes you can find 10 trades/day. I’m not sure about you, but I don’t have the luck of finding a new stock that I am convinced enough to invest in on a daily basis, let alone ten such stocks.

3. The game also assumes the balance of the account is untouched, and there are no inflows or outflows.

Application of These Lessons

But that’s enough on stipulations. In terms of takeaways from this game and the Kelly Criterion on the whole, I would say a few things:

1. The biggest takeaway that you should have from the Kelly Criterion is that if you do not have an edge, you should not bet. With a 50% probability of winning and losing, the optimal bet size is zero.

2. Second, over-betting is possible. See the below excerpt from a great piece on Probability and Payoffs by Michael Mauboussin and Dan Callahan at Counterpoint Global

3. Because overbetting is possible, especially when using the Kelly Criterion for investing, there are valid arguments to utilize 1/2 or 1/4 Kelly to lower the volatility. You get roughly 75% of the growth at half the vol by using 1/2 Kelly.

   

4. There is evidence that Buffett, Soros, and many other great investors, whether knowingly, or it coming to it from other means, seem to apply the Kelly Criterion to their stock market practices. You probably should too.

   

5. It’s also important to recognize that we cannot truly know what the odds of winning are when placing a bet, or making an investment in our case.
   
   What we can do, and what I attempt to do in my stock picking, is to raise your bar for what is investable, through whatever means. For a Buffett-Munger style value investor, maybe this means your margin of safety is 30% instead of 15%. For you, maybe this means 50 or 100 hours of due diligence.

6. Part of the utility in applying the Kelly Criterion to stock picking is not just finding the right sizing for a single position, but coming up with a comparative figure that you can use to size multiple positions within a portfolio. Part of good portfolio management is managing opportunity cost.
   
   For every dollar you allocate to NVDA, that’s a dollar you could allocate to another company. Is NVDA better than every stock in the market? Probably not, but you can’t do due diligence on every stock in the market. But you can compare NVDA to other companies you have done the work on, and determine based on your relative view on each of those, how you should allocate to that.
   
   This is part of why active management is such a tough game, there are countless opportunities out there, but you have to do the work in turning over those stones.

   
   As you turn them over, you should try and assign a value, a score, anything based on your confidence in them, catalog it, and use that aggregated information to build a diversified portfolio of positive EV bets.

7. Like I’ve mentioned above, it’s unlikely that you will be able to accurately forecast your odds of winning, and that’s not really the point of this exercise anyways. The point is that you should go through the thought process of trying to forecast.
   
   This practice will force you to assign measurable figures for what used to be your “gut feeling” about a stock. It will help you detect patterns between the investments you gravitate towards and why.

   It will also be useful to look back on and ask, “where did I go wrong?” or, “What assumptions did I underwrite which didn’t come to pass?”
   
   As you forecast, review your forecasts, and make more for the future, you should get better at it. It should serve to help bring words to feelings and make you more confident in your investments and in your thought processes.

Quotes & Anecdotes on Portfolio Construction

To end this article, I wanted to give you some anecdotes and statistics which lend credibility to the difficulty as well as the importance of the art that is portfolio construction.

"I have two basic rules about winning in trading as well as in life: ‘If you don't bet, you can't win. If you lose all your chips, you can't bet,’” from Larry Hite’s interview in Market Wizards.

“All you need for a lifetime of successful investing is a few big winners, and the pluses from those will overwhelm the minuses from the stocks that don't work out,” from Peter Lynch.

Buffett: "Diversification is protection against ignorance. It makes little sense if you know what you are doing."

Munger's version: "The idea of excessive diversification is madness."

Druckenmiller has echoed the old Carnegie line: “put your eggs in one basket and watch it closely.”

Modern portfolio theory suggests that holding 20 to 30 stocks from different industries captures most diversification benefits. A portfolio of 25 well-diversified stocks can eliminate about 90% of unsystematic risk, according to empirical research.

It’s important that I emphasize “well-diversified” here. Twenty AI infrastructure bets is still one bet.

Conclusions

Taking in all of these quotes, books I’ve read, as my lessons from my own experiences in the stock market, I’m left with a few concrete takeaways that apply to all investors:

1. The first and most important rule of capital allocation is to not lose money.

2. If you’re gonna bet big, you need strong conviction, and you better be right.

3. It’s very hard to be right, and to consistently be right over many investment cycles.

4. If you don’t have the time, energy, or temperament for this game, you should stick to ETFs.

5. Time in the market does beat timing the market. There’s a reason I like to go long-dated with the options plays I do, and it’s because while I may have conviction a stock is cheap, I don’t have conviction that a certain event will make the market agree with me.

6. Valuation matters, and don’t listen to anyone who tells you otherwise.

   

   I’m not saying there are no investments that work at 40x earnings, but I am saying that your hit rate is bound to be lower the higher the earnings multiple of the stocks you invest in.

   

At the end of the day, you have to decide what cross to bear. Concentration and potentially higher returns with higher volatility? Or diversification, and its tradeoffs? In my own portfolios, I opt for concentration and have to be comfortable with higher volatility as a consequence of that choice.

Leave a comment

Enjoyed this article? Consider subscribing, leaving a comment, or check out my most popular posts:

The author of Dune, Frank Herbert, was something of a sci-fi visionary, at least to me. One of the main concepts within Dune’s world building is that there was a war between humans and what he refers to as “thinking machines.”

The Butlerian Jihad, this war between humans and thinking machines, had drastic effects in the Dune universe. The Orange Catholic Bible added a commandment:

“Thou shalt not make a machine in the likeness of a man’s mind”

The knock-on effects of the Butlerian Jihad are what I find more interesting. (I know this is some nerdy shit, but I’m getting to the point)

The ban covered any machine that “thought” like a human, down to simple calculators. In a civilization where space travel is normal, this poses many problems.

Dune’s solution to these complex problems is to elevate the human mind:

“It forced human minds to develop. Schools were started to train human talents.”

In today’s world, I see the very early signs of humans outsourcing our thinking to machines. In my later years of college, namely 2024 and 2025, when LLM capabilities were vastly improved, many of my peers had a mindset of “let’s just have Chat do it.”

I cannot even imagine how high schoolers or kids entering college today rely on AI.

But this phenomenon is not localized to just students. I see countless examples of people using LLMs for tasks (or thoughts) they could have done (or thought) themselves.

So What? What’s The Big Deal?

When this topic gets broached, I typically hear the response of so what? We are spending trillions of dollars developing this technology; it’s happening and there’s no stopping it.

I don’t want to stop the progress of AI. I want it to rapidly improve and for people to use it to cure diseases, increase productivity, and help humanity. My fear is what happens to us.

I fear that if someone starts using Claude at 11, and is able to ask it for help whenever they get stuck, what will they ultimately learn? And for those of us who have gone through school already, how much of our knowledge/intelligence will atrophy from rampant AI use? How much already has?

How do we protect ourselves from this?

I have not been immune to AI reliance myself. I constantly ask Claude or ChatGPT for EPS projections, or historical figures, when I could easily run the numbers, pull up the 10-K, or find the estimates.

But I don’t. Is it laziness? Ease of access? Probably both. But over the long run, how will I gain more experience or be able to identify patterns between companies or situations?

If my first resort is to ask Claude, I remove a layer of friction where I am forced to think of why something may be occurring.

When I had to write a free-response paper in ninth grade English class, I had to think back on the characters, prose, setting, motivations, and plot of the novel I was answering a response on.

My brain had to juggle all of those, while simultaneously crafting a coherent response to a sophisticated question posed by my teacher. All within 38 minutes.

I’m ashamed to admit that my brain has not had to work that hard in recent years. Between scrolling Twitter, watching YouTube, listening to podcasts, my brain rarely gets a quiet moment.

My brain takes in more information weekly than I probably took in within a single month of high school. But it doesn’t utilize it. Or fully retain it. It just takes in information, then makes more room for whatever podcast comes up next, or whatever WSJ headline catches my attention next.

Maybe I am alone in this experience, and everyone else is mindful of their use of this technology. From what I have seen personally, I am not.

There’s No Butlerian Jihad Coming

We’re (probably) too early for a war against these thinking machines. As AI proliferates throughout our economy and society, it is undoubtedly going to drive massive benefits, at least in the near term.

We may never have a war between humans and AI, and that’s not why I wrote this article. I think the more useful takeaway from Herbert’s Dune is that instead of stalling AI progress, we need to stimulate human progress.

There’s not going to be a ban on thinking machines, at least not for a long time. To improve your thinking, you will have to impose these bans for yourself. Restrict yourself from using a calculator, and work on your mental math. Write, or journal, and sit with your own ideas without asking an LLM to organize them. Go through analog processes, build out your own Excel models, and dig into assumptions by yourself.

Ultimately, I think the unlock will be using these LLMs to create hyper-personalized learning modules that you can use to train your own mind, and become a Mentat of sorts.

Using AI To Think Again

As I was brainstorming this article, the realization that I was passively consuming massive amounts of information, yet not actively utilizing it, was pretty scary.

After all, what’s the point of listening to a podcast, reading a book, or watching a video if you can’t even remember or utilize the information from it?

Information Utilization

For that reason, a good starting use case for AI is a journal prompting tool. I think inverting the relationship, and asking AI to prompt you to create something is useful because a lot of the time people go to these tools so they can put low effort in, and get something out that looks half-decent.

When you invert the relationship this way, and have AI give you something to think deeper on, it can spark more creative thoughts. Now of course, journal prompts have been around forever, but what if you had one personalized to whatever you’re experiencing in your day-to-day life? One that constantly updates with new context about you and knows what you are trying to accomplish.

It can ask you something related to whatever book you’re currently reading, how the first day of a new job is going, how a project you’re working on is developing, etc.

Give Yourself Homework

Another use case that I have enjoyed recently is using AI to give myself homework. To the employed, this may seem insane. But for me, it has been really helpful. You see, in recent years I feel like my mental math and recall have declined somewhat.

In fact, I feel like I felt sharpest when I was in the middle of high school, learning calculus, and having weekly quizzes that I dreaded. I’ve realized, especially when I had to study Spanish in high school and college, that if you don’t use something, you’ll lose it.

This goes for all types of knowledge in my experience, and how can that not be scary in a time where AI can do so much for you?

What I’ve done so far, and I hope to expand this protocol significantly:

1. Every day, using a scheduled task, Claude creates a quick math worksheet for me to do, covering addition, subtraction, multiplication, exponents, fractions, percentages, as well as financial ratios.

2. Every day, Claude finds a Fermi estimate problem for me to approximate. The goal with a Fermi estimate is to get within an order of magnitude of the real answer. For example, I am tasked with approximating the number of licensed real estate agents in the US, and I go through a series of logical assumptions with the goal of getting close to the real answer.

1. Reading/Recall exercises: I have Claude find me an article/historical passage to read, and then do the 10/24/7 Method with it, where I answer questions 10 minutes after reading, 24 hours later, and 7 days later.

Other use cases I’ve been thinking of:

1. Curriculum builder for new interests: let’s say you get into writing, carpentry, or meditation. You can ask Claude to find you a comprehensive list of resources to study. You can discover new authors, YouTubers, podcasts, whatever it may be.

2. Reading partner: maybe you don’t need to do full-on book reports, but you can have an LLM pose or find existing questions about books as you read them, prompting deeper thought from you.

Go Analog When Possible

My main goal with these protocols is to shift the relationship between myself and AI. Initially, AI was useful for bouncing ideas off of, building simple financial models, and organizing my writing.

What I found after using it pretty extensively was that these actions, however simple or mundane they may be, grounded me in the work, and made me think more deeply. When I wrote the discount rate, or projected out future growth, I was more tied to that number. I had to assess it more closely and determine if it was reasonable.

When Claude/ChatGPT did it, I just assumed that what they gave me was a reasonable figure. I didn’t question it, and a lot of the time I’m sure they were reasonable answers. But I wasn't tied to that figure. I couldn't defend it.

When I had an LLM organize my writing, I lost the time where I would look over my thoughts with fresh eyes. I removed the opportunity for myself to reassess what I wrote.

There is a time and a place to use these tools, but there’s also a time and a place to think deeply on your own. To ask why something is the way it is, and to allow for that moment of friction where you don’t know the answer to a question. That’s why you should put the tools down every once in a while, and go analog.

This is not projected to stop, as these companies, for the time being, are adamant about continuing to acquire compute at high costs. The problem? The market wants to see clearer returns on investment.

Taking a look at the valuations for these hyperscalers, you can see they’ve taken a hit from their historical levels. Specifically, the group of AMZN, MSFT, and META have taken the largest hits, which you can see in the next graph. Apple, despite facing increasing memory costs in all of their hardware, is helped by the fact that it has not made significant capital expenditures in this AI spending wave.

What’s the Point?

Up until this point, this article has been mostly descriptive, and not predictive of what I think will happen. I wanted to provide an unbiased overview of the current state of the market before getting into my beliefs about these companies.

With that being said, I have a few key beliefs about the current state of things.

1. These semiconductor valuations, as well as market sentiment do not make sense. To explain, let’s look at those capex projections again.

   

   BofA’s latest report has capex for GOOGL, META, and AWS at over $1.5 trillion for the next 2 years, not including 2026 spend. That does not include MSFT. Yet, NVDA, which you could logically presume to be a major winner from this level of capital spending trades at 15x FWD earnings. Alongside that, the memory companies which have gone parabolic over the past year, still trade at single digit PE ratios.
   
   The only way that those valuation levels make sense is that these capex projections are totally off, there is some silver bullet coming for memory utilization in LLMs, or that somehow the capital decision making of these 85-90% margin memory companies will erase the massive returns they are earning now. I find those unlikely.
   

2. At the same time, there is this belief within the market that hyperscalers are willingly overspending, and that there is no line of sight to ROIC on this massive capex. I'm against that idea for a simple reason: these companies have highly intelligent founders and capital allocators at the helm.
   
   They have a vested interest to make high returns on their capital decisions, and have huge financial outcomes tied to making that happen. The idea that they would willingly misallocate billions of dollars is insane to me.

   In META's case specifically, I think the Metaverse pivot soured investors and gave Zuckerberg a reputation for lighting capital on fire. That's exactly why the setup is compelling, as if and when they show clear returns on this spend, the sentiment shift and re-rate could be violent, similar to Google’s sentiment shift in mid-late 2025.
   
   Additionally, these companies have multiple paths to gaining ROIC on this capex: selling compute to frontier labs, running their own models, running other models locally, replacing large swaths of the workforce via the improved productivity of AI-enabled employees, along with others.
   

3. The market is pricing the outcome of AI as zero-sum. It feels as though there are no days where the hyperscaler/Mag 7 factor can be up while the semis factor is up.
   
   The sentiment is that semis spend is unsustainable and that the hyperscalers are going to stop cutting checks and these semiconductor stocks will plummet. There is no middle ground or belief that this demand for semiconductor stocks could be a secular shift and not us spending at cyclical peak levels.
   
   While I do not predict that we will spend trillions of dollars on datacenter buildout in perpetuity, I do believe there is some middle ground where the demand for compute grows at a high rate, while big spenders see a ROIC.

Actionable Takeaways

I believe there are a few key takeaways from this discussion. Firstly, if you are in the camp that we are at the peak of a massive overspend on compute, these hyperscalers can still be good investments. Many of these companies, namely Meta and Amazon, would re-rate massively if there were signals of capital discipline. It’s not as though their spending has impaired their core businesses, and while losses of hundreds of billions of dollars in the worst case would obviously hurt, these companies could earn that in operating cash flow within a few years.

Second, if you are a believer in this middle ground, NVDA at 15x forward earnings seems pretty attractive. Not to mention the memory companies that are printing cash right now. Samsung just overtook Nvidia as the most profitable company in the world, with ~$58.5 billion in operating profit for Q2 2026.

Lastly, I want to emphasize that I was not a market participant during the 2000s dot com bubble, and I have seen many claim that those who lost money were foolish or were making terrible investments. When we look back at financial crises, we see the real figures, the real EPS that happened, and not the forward earnings projections made by the market. In hindsight, it is always easy to call something overvalued or investors stupid for thinking pets(dot)com was a good investment.

The reality is that I’m sure a lot of sound, rational market participants lost money in 2000, and that things probably looked logical in a forward valuation sense. While I am not an AI bear, I do see possibilities for this shaping to be another situation similar to that, and for that reason I have shifted many of my investments to these mag-7 companies which have the benefit of having sustainable core businesses that are positively affected by AI.

I think one of the key things to watch is how concentrated the compute becomes. As SpaceX, and potentially Meta sell their “excess compute” to Anthropic, the “key man” risk increases, and the demand for compute becomes reliant on few companies.

It is also important to note that many of the current subscription models do not seem to be sustainable for Anthropic/OpenAI, as they continue to try to shift certain models to token-based billing, despite the consumers distaste for that model. The level of subsidization of these models remains a key discussion point for bears and bulls, but without internal company data it is pointless to argue over.

My hope is that these companies IPO sooner rather than later, so that we can have line of sight into these factors, but it seems the SpaceX IPO, Google equity raise along with the coming SK Hynix ADR, may be sucking the air out of the room for some time.

SpaceX is set to IPO tomorrow in what will be the biggest IPO in history, at least for now. Anthropic and OpenAI appear to be close behind.

The numbers: $135 per share, 556.6 million shares offered, a $75 billion raise at a $1.77 trillion valuation. Only about 4% of the company will actually trade.

With a float that small, and retail interest of around $70b before the open, I expect a lot of volatility. Reported total demand sits somewhere between $150b and $250b, depending on who you ask. My guess is it trades well above the $135 listing price.

For scale: the previous record holder was Saudi Aramco, which raised about $29b in 2019. SpaceX is raising roughly 2.5x that in one shot.

The Numbers

In 2025, SpaceX did $18.7b of revenue and lost $4.9b. So the company is coming public at nearly 100x trailing sales while losing money. There is no P/E multiple because there is no E. If you want to be generous and use 2025 adjusted EBITDA of $6.6b instead, you’re still paying about 265x. Back of the envelope: if SpaceX handed shareholders every dollar of that EBITDA forever, it would take more than two and a half centuries to earn back the purchase price.

Of course, the market is forward looking. So we should be too.

Where the Revenue Comes From

Two things matter here: Starlink and, as of about three weeks ago, datacenters.

Starlink is the real business. The connectivity segment did $11.4b of revenue in 2025, up roughly 50% year over year, with $4.4b of operating profit. Profitable, recurring, growing. If this IPO were just Starlink at a reasonable price, I’d be interested.

The datacenter side is what’s carrying the valuation. The S-1 disclosed that Anthropic agreed to pay $1.25b per month through May 2029 to rent the entire output of the Colossus datacenters. Then last week, days before the IPO, Google signed on at $920m per month through June 2029 for roughly 110,000 NVIDIA GPUs. That is about $26b of annualized compute revenue from two customers, signed within weeks of the listing.

SpaceX announced Q1 2026 revenue of $4.7b. Add the Anthropic and Google deals, which would bring in roughly $13b over the rest of the year, then layer in the base business. You get to roughly $33b of 2026 revenue.

With those assumptions, the IPO is priced at about 53x current year sales. Again: sales, not earnings.

Paths for Growth (and Why the Growth Is Fragile)

76% year-over-year revenue growth looks great, but most of it comes from the Anthropic and Google deals. The question is whether SpaceX can keep building compute at this pace and rent it out at top-of-market rates while everyone else is also trying to sell GPU capacity.

Based on the S-1, SpaceX could have 300-500MW of compute online and leasable by mid-2027. At current rates, that could bring in $500m to $1b per month. In the rosiest scenario, call it $6b of incremental revenue in 2027. The problem is that NVIDIA will also be shipping GPUs to CoreWeave and the hyperscalers over that same period. Supply goes up.

And then there’s concentration. Two customers account for essentially all of the incremental growth, and the S-1 names both of them as competitors.

Comparison to the Megacaps, Specifically $META

This is the part that gets lost in the headlines. SpaceX is listing above Meta’s current market cap. Meta does roughly $200b in revenue at around 50% EBITDA margins and trades around 18x forward earnings. SpaceX did $18.7b and lost money.

You’re being asked to pay more than Meta’s price for about a tenth of the revenue and none of the profit. Both companies are spending enormous sums on AI. Only one of them generates the cash flow to pay for it internally.

What You Have to Believe at $135

Let’s reverse engineer the price. Say we assume a 2031 P/S multiple of 15x, which would still be a big premium to most companies, and a 15% annual return over the next five years. SpaceX would have to maintain roughly 50% revenue growth every single year, reaching about $240b in revenue in 2031, just to make those assumptions work.

And that is only for a 15% annual return. My guess is plenty of retail buyers tomorrow are mentally underwriting something much higher than that.

The Index Inclusion Game

The float is small on purpose, and the whole structure looks built to get into the indices quickly. Under new Nasdaq rules effective May 1, a company of this size can enter the Nasdaq 100 within about 15 trading days of listing instead of waiting a full quarter. FTSE Russell has cut its window to 5 days. BNP Paribas estimates roughly $8b of forced passive buying in the first month from Nasdaq 100 inclusion alone, and something like $30b of total passive demand across the index families. If you own a Nasdaq 100 fund in your portfolio, you’ll own SpaceX within weeks whether you wanted to or not.

The xAI Escalator

It’s worth being clear on who gets liquidity tomorrow. In 2022, a group of co-investors helped Musk take Twitter private at $44b. By 2024, some of those stakes were marked down roughly 70%.

Then xAI absorbed X in March 2025 at an $80b xAI valuation, converting their underwater social media bet into hot AI startup equity. Then SpaceX bought xAI in February 2026 in an all-stock deal that valued xAI at $250b.

Tomorrow, those SpaceX shares start the clock toward being publicly tradeable.

So the 2022 Twitter co-investors rode an escalator from a deeply impaired social media position to liquid stock in a $1.77 trillion public company. This IPO is the top floor.

Retail FOMO Is the Product

30% of the offering is earmarked for retail, roughly triple the normal allocation for a deal this size, distributed through Fidelity, Robinhood, and the big retail channels. The headlines have done the rest. Demand of $100b against $75b offered.

When a deal is built to maximize retail participation and the float is deliberately thin, you should ask who the structure serves.

Yes, It Can Obviously Pop

To be fair to the bull case: thin float, oversubscription, forced index buying, a retail base that won’t sell on day one, and the Musk premium. This thing can absolutely rip in the short term. I would not be surprised to see it trade well above $135 out of the gate. Some are already betting the day-one close lands above $2 trillion.

The Near Term Selling Pressure

Additionally, the lockup structure rewards the pop with supply.

After the first earnings report as a public company, insiders can sell 20% of their eligible shares. If the stock is also trading at least 30% above the IPO price for five of ten trading days, they unlock an additional 10% on top. Then 7% tranches release at 70, 90, 105, 120, and 135 days. Another 28% unlocks after Q3 earnings, and the remainder frees up at the standard 180 days. Musk himself is locked for 366 days, which matters. But everyone else gets out early, and faster if the stock goes up.

The Trade

Most of SpaceX’s growth, at least in this version of the story, comes from the compute deals. Compute supply is growing quickly, and that could pressure rates. Maybe demand keeps growing fast enough to absorb it. But if that’s the bet, NVDA looks like a cleaner way to express it than SPCX.

Read more

(Most) every stock you own has an entire derivatives market sitting on top of it, constantly repricing the odds of every possible outcome. Hedge funds, institutional desks, professional traders. All of them are putting real money behind their convictions every single day.

That activity gets compressed into one number: implied volatility. And most equity investors completely ignore it.

Here’s how to read it.

What implied volatility actually is

Implied volatility (IV) is the market’s forward-looking estimate of how much a stock is expected to move, expressed as an annualized percentage.

It’s called “implied” because it’s backed out of the price of an option. Historical volatility measures how much a stock has moved. IV reflects how much the market expects it to move. When option prices are expensive, IV is high. When they’re cheap, IV is low.

Think of it as a fear gauge for a specific stock. A stock with 20% IV is expected to be relatively calm. A stock with 60% IV is expected to swing hard.

But on its own, that annualized percentage is tough to intuit. So here’s how you actually make it useful.

Turning IV into an expected move

The formula is simple:

Expected move = IV ÷ √(trading days in period)

For a daily move, you can simplify: divide IV by √252 (trading days in a year), which is roughly 16. That’s the shortcut worth memorizing.

If a stock has IV of 32%, the market is pricing in roughly a 2% daily move (32 ÷ 16). If IV is 48%, that’s 3% a day.

For a weekly move, divide by √52. For a monthly move, divide by √12.

This gives you a one standard deviation move, meaning the market prices roughly a 68% probability that the stock stays within that range. A 2σ move (which covers ~95% of outcomes) is just double that.

Important: this isn’t a prediction. It’s a probability-weighted range. The market isn’t saying the stock will move 3% tomorrow. It’s saying that given current option pricing, a 3% move in either direction wouldn’t be surprising.

What happens when IV goes Parabolic: GameStop in January of 2021

In late January 2021, GameStop’s implied volatility hit 1,000%. Run that through the formula: 1,000% ÷ 16 = a 62.5% expected daily move. The options market was saying GME could swing 62% in a single session and that would be normal.

Obviously absurd. And it was a massive opportunity for anyone selling premium.

IV almost always overstates the actual move. Options are insurance, and insurance is almost always priced above the expected loss. When IV goes parabolic, that gap between what’s priced in and what actually happens gets enormous.

If you sold options when GME IV was at 800–1,000%, you were collecting premiums that assumed 50–60% daily moves. Realized vol eventually collapsed well below implied, and sellers got paid handsomely as IV crushed back down over the following weeks.

Why this matters right now: SpaceX.

SpaceX filed its S-1 with the SEC on May 20, 2026, confirming plans to list on Nasdaq under the ticker SPCX. Reports point to a debut as early as June 12. This is going to be the largest IPO in history and it’s not even close.

The numbers from the filing are wild. Revenue hit $18.7 billion in 2025, up 33% year over year. Starlink is the engine here: 10.3 million subscribers across 164 countries, $11.4 billion in segment revenue on its own. They also disclosed a $1.25 billion per month cloud compute deal with Anthropic running through May 2029, which tells you something about the scale of AI infrastructure they’re building at their Colossus data centers in Tennessee.

But SpaceX also lost $4.9 billion in 2025 and $4.3 billion in just Q1 2026. Capital expenditures nearly doubled to $20.7 billion as xAI (which SpaceX acquired in February 2026) burns cash at an extraordinary rate. The company carries $29.1 billion in long-term debt. The AI segment alone posted a $6.35 billion operating loss in 2025.

The self-assessed valuation in the prospectus is $1.25 trillion. Media reports suggest the final pricing could land anywhere between $1.75 trillion and $2 trillion. At the midpoint, that’s roughly 94x trailing revenue for a company that is deeply unprofitable on a GAAP basis. For context, Tesla’s price-to-sales ratio at its 2010 IPO was a fraction of that.

Now here’s the part that actually matters for this article. The float is expected to be roughly 5% of shares outstanding. Musk has earmarked 30% of the offering for retail investors, which is triple the typical IPO allocation. Demand is expected to be 10–20x oversubscribed. And under Nasdaq’s fast-track entry rules, SPCX would qualify for Nasdaq-100 inclusion after just 15 trading days, triggering forced buying from every index fund and ETF tracking it.

Think about what happens to IV on day one of SpaceX options trading. You’ve got a company valued north of a trillion dollars, massive retail interest, a tiny float, forced index buying on a compressed timeline, and the Elon Musk premium baked in. IV will almost certainly be extreme.

I think we could see something approaching the levels from the GME/AMC meme stock frenzy of 2021.

META: What the Market is Pricing right now

Let me ground this in a real example with a stock I own.

META is trading at $607.32 as I write this. The $610 call expiring June 26 (about 32 days out) has an implied volatility of 30.24% and a delta of 0.5343.

Plugging into the formula:

$607.32 × 0.3024 × √(32 / 365) ≈ ±$54.35

That puts the market’s 1σ expected range at roughly $553 to $662 by June 26.

That’s the cone below:

A few things worth noting.

This range reflects the aggregate positioning of every participant pricing META options right now. It’s not one analyst’s opinion. It’s the market’s collective bet, backed by real money.

IV of 30.24% is moderate for META. The IV rank is sitting around 23–25, meaning current implied volatility is near the low end of its range over the past year. The market is not pricing in much event risk over the next month, which makes sense because META’s next earnings report isn’t until July 28. Between now and June 26, there’s no major catalyst on the calendar.

The range here is symmetric, but skew (the difference in IV between puts and calls at equal distances from the current price) can shift it. If put skew is elevated, it pushes the downside boundary wider, which means the market is more worried about a drawdown than a rally.

Delta: the market’s implied probability of hitting a level

The cone gives you the range of what a stock could do. Delta gives you the odds of hitting any specific price within that range, or beyond it.

Delta is a Greek that options traders use to measure how much an option’s price moves relative to the stock. But it also works as a rough proxy for the market-implied probability that a stock reaches a given strike by expiration.

A call with a delta of 0.50 sits right at the money. The market is pricing roughly a 50/50 chance the stock finishes above that strike. As you move the strike higher, delta falls, reflecting lower odds of the stock getting there in time.

Back to META. The $610 call expiring June 26 has a delta of 0.5343. The market is pricing about a 53% chance META is above $610 in 32 days. Basically a coin flip, which makes sense since the strike is right near the current price.

Now think about a $700 call on the same expiry. You’d be looking at a delta around 0.10. The market would be saying there’s roughly a 1-in-10 chance META rallies 15% in a month. Not impossible. But the options market, with real money behind it, is pricing that as a low-probability outcome.

This is useful even if you never touch an option. It’s helpful to play around with options, looking at different deltas at different time horizons to understand what probability the market is ascribing to certain outcomes.

For example, I am a META bull, and my current price target for 2027 is $1000. If we take a look at $1000 December 17, 2027 calls, the market is only giving a ~25% chance of META hitting that level.

Put volume: when the market is actively hedging

IV tells you how much movement is priced in. Delta tells you the odds of hitting a specific level. But there’s a third signal worth watching: put volume, or how aggressively market participants are buying downside protection.

In early April, SPY put volume surged past 9 million contracts. That’s the third-largest single-day spike on record. Institutions, funds, and sophisticated traders paying real money to hedge against a drawdown.

When you see that kind of spike, it doesn't mean a crash is coming. It means the people who manage the most money in the world are collectively saying: the risk/reward of not hedging is worse than the cost of insurance. That's a signal worth paying attention to, even if you never buy a put in your life.

How to use this as an equity investor

You don’t need to trade options. But knowing this range changes how you think about your position.

Position sizing. If you’re long META and the market is pricing a ±12% move over the next month, that’s $70/share of expected volatility. Are you comfortable with that level of volatility?

Setting price targets. If you have a 12-month price target well inside the 1σ annual range, the market considers that part of the base case.

Evaluating conviction. High IV means options are expensive, which means the market is uncertain.

Timing entries. IV tends to mean-revert. Buying LEAPS on a stock when IV is at a multi-year low means you’re entering when the market is pricing in low volatility. That can be opportunity, as you have two outs to profits: share price appreciation and implied volatility expansion.

Where to find this data

All of this is freely available.

Any Modern Brokerage — Robinhood, IBKR, Fidelity, etc. (their options chains)

Market Chameleon — IV percentile rank, historical IV vs. realized

Other Thoughts & Updates

I’m drafting an Eli Lilly ($LLY) write up that I plan to post in the next few weeks. I think it’s trading at a fair multiple, right in the GARP-range and has a massive tailwind if Retatrutide (reta) gets FDA approval. If Reta is legally cleared as a biological compound (big differentiator) I think $1150-$1200 is likely as that will give them a much longer exclusivity window. I personally think that based on their EPS estimates and forecasted growth, $1500 by ~mid 2028 seems feasible. No position yet but I plan to open call spreads soon.

Disclaimer: This is not financial advice, do your own due diligence please.

But for some reason, there remains alpha within one of the largest companies in the world, and it seems to be available on a regular basis. I’m talking about Meta, which for some time now has traded as an AI loser, despite showing enormous returns, expecting massive forward earnings growth, and maintaining a reasonable valuation.

Why is this Opportunity Available?

With every capex increase over the past few years, Meta stock has traded down. People seem to have PTSD from the Metaverse/VR spending cycles of ‘22. The datacenter capex is substantially different for multiple reasons which I will get into here.

Firstly, datacenters take ~2 yrs to get online, and Meta spent $28b in DC-related capex in 2023, so we are starting to see the first signs of AI-improved ad impressions + pricing. Anecdotally as well, Instagram’s Reels algorithm is by far the best at keeping my attention, as well as others. It is not uncommon that I will open Instagram to find 5+ reels from some friends.

Additionally, their ads are by far the best of any social media platform. X’s video ads are pure slop, and the text posts are completely nonsensical. YouTube’s advertising is pretty good, but I have only ever bought something from an ad on Instagram personally.

Secondly in regards to the AI spend, xAI (and neoclouds) have recently shown that there are multiple paths to monetizing compute, with Zuckerberg himself saying this in Q3 of 2025:

Like almost every week, people come to us from outside the company asking us to stand up an API service or asking if we have different compute that they could get from us. And we haven’t done that yet, but obviously if you got to a point where you overbuilt, you could have that as an option.”

Thirdly, they have massive distribution channels with 3B MAUs on Instagram, and even more across WhatsApp and Facebook. If models are going to be commoditized, and largely homogenous across companies, than distribution channels to customers is going to be a major differentiator.

The Increasing Diversification of Meta’s Business

Google has grown to become such a large company mostly on the back of their core business, but also their “other bets” that have grown into massive divisions. I believe Meta is in the early innings of their own version of “other bets” and I see clear paths to each of these being substantial divisions in their own right.

• Developing their own OpenClaw with MuseSpark

  

• Meta Raybans - AI infused glasses

  

• Massive amounts of compute for the exponentially increasing demand

• VR, AR, Meta AI assistant

So What?

It’s my belief that Meta presents a similar opportunity to Google in early 2025. It’s a company that everyone knows, understands, yet they are missing the forest for the trees. Yes, Zuck is dumping boatloads of money into compute - but it’s already seeing returns. And guess what, if they end up not having a good use for the multiple GWs they plan to bring online, I’m sure Anthropic and OpenAI would be very interested.

For some reason, you can get one of the best advertising businesses in the world for 18x forward EPS. Also it’s growing revenues at 30%/yr. Oh, and they’ve also shown a willingness to buyback shares and get lean when the stock price tumbles. I think once Meta’s products, or AI-related returns meaningfully show up in the numbers, they are due for their own ‘Gemini 3 Moment’ like Google had last November.

In my opinion, once this happens they are very clearly a 30x PE company, with consensus EPS estimates around $35 for 2027, this quickly turns from a $600 stock into a $1000+ stock as soon as next year. Obviously to cause this re-rating, there would likely be a substantial EPS beat, so the stock could well exceed $1000.

The funny part about this is that for Meta, and for the S&P as a whole, 30x earnings has been pretty reasonable. Their average PE ratio since IPO is 29.4x, and the S&P’s current forward PE ratio is 20.9x. Meta is trading like a peer amongst the S&P despite its exceptional results, and I believe this is a temporary mispricing similar to the drawdown in ‘22.

I am personally long Meta, and increasing my position such that it is my largest by far.

Disclaimer: This is not financial advice, do your own due diligence please.

The stock market works the same way. In the dot com bubble, people were (correctly) sure that the Internet would change the world. It did. Yet if you invested in the wrong companies (or at the wrong prices) you could have lost a decade’s worth of returns, if not more. See INTC and others below:

INTC: Not adjusted for dividends, if you bought at the higher levels in 1999-2000, you are still down to this day, two and a half decades later.

Cisco just recently reached the same levels seen during the peak of the internet bubble.

It took ORCL 14 years to reach the levels seen in 2000.

These are the companies that survived the dot-com bubble too, there are countless others that didn’t. We are likely to see companies that don’t survive this cycle either.

What’s the Point?

I’m not mentioning these companies because I foresee some massive downturn in the market. In fact, I am very bullish on AI, and many of the largest positions in my portfolio, like EWY, hinge on the continued performance and demand for AI compute.

I am writing this as a reminder to myself and others to not get carried away with the irrational exuberance that is often seen at the top of bubbles.

Allbirds (formerly a shoe company) recently just pivoted into selling AI compute. The stock is up 300% today (sorry 700% now). These are not normal market moves, and while it could be justified, I think it is rational to flag these as at least mildly concerning developments.

Regardless of whether or not you believe that AI is the real deal, and that it is going to deliver massive returns, you should still be rational with portfolio allocation, and in that effort I have some recommendations.

Actionable Advice

Recently, I have been adding more defensive positions to my portfolio, and I think this makes sense for many reasons. Firstly, they act to lower your Beta, or correlation with the broader market. Secondly, they are more resilient in tough times (like if the Iran War were to re-escalate.)

Thirdly, they help you position for the binary-like outcomes that come with the development of AI. To explain, let’s say AI really works out and replaces tons of white collar workers, leading to mass unemployment. Consumer spending would theoretically fall drastically in this outcome. Consumers would still spend on staples, as they have to in order to survive.

On the flip side, let’s say AI capex falls through. Companies don’t see ROI, LLMs are good, but don’t deliver the economic returns that the massive capex spending necessitates. Humans will still be working, unemployment will go down as companies see that AI isn’t going to replace workers the way they expected it to. Consumer spending goes up in this scenario, and defensive stocks would be fine, if not up more than the broader market.

In conjunction with being long AI, a defensive positioning makes sense to me. The S&P 500 is already ~35% mega-cap tech, so a "neutral" broad-market allocation already carries significant AI exposure. A rough framework: 25% direct AI, 25–35% defensives, with the remainder in the broader market. Obviously this is broad advice and should be adjusted based on age, risk tolerance, etc.

AI works out —> AI positions up, broader market up, defensive stocks flat/underperform

AI doesn’t work out —> AI positions down, broader market down, defensive stocks flat/down less than broader market.

Median outcome (my assumption) —> AI positions up, broader market up, defensive stocks up/flat

Obviously the median outcome is quite vague, but I picture it being a scenario in which AI is successful, delivers returns, but doesn’t quite change the world over the next 3 years. AGI doesn’t come (yet) and you want to be positioned in the picks and shovels (semis) rather than the LLMs themselves.

Concluding Thoughts

This is all to say I don’t really know what’s going to happen. No one does. I would be hesitant to take financial (or any) advice from anyone speaking in certainties on AI. While there are many reasons to be optimistic, there are also many reasons to be cautious. Oftentimes, the truth lies between the two extremes and it is my belief that it is best to be positioned to benefit from the widest range of outcomes. The music is still playing, and it may be for some time. Just be ready for whatever comes next.

Disclaimer: This is not financial advice, do your own due diligence please.

Despite Q1 FY26 guidance of $78 billion, an over 70% YoY growth, NVDA’s forward price-to-earnings multiple has contracted to ~21.4x.

This is significantly below its average of ~39.1x over the past five years. Obviously there have been macroeconomic headwinds that are partially at fault here, but I think NVDA’s fall from higher valuations is interesting.

Looking forward, I am still bullish on GPU demand as demand for compute will only grow from here:

In roughly a year, weekly token usage has gone from ~2.18T tokens to ~22.6T tokens. I only expect this to grow in the near term, especially as large companies are pressured to reduce headcount — inference demand will grow as companies downsize.

This is something I want to devote a full post to in the future, but in looking at the future of white collar work I think a few things are becoming clear:

1. Junior/early career roles are largely disappearing

2. Mid-level employees will become vastly more efficient or will be part of the layoffs

3. Economic rents will shift from humans to models/applications.

The third point is probably the most defensible. When you pay an employee a market rate, that rate has a biological floor — it has to cover housing, food, healthcare, and everything else a human needs to survive. That floor props up labor costs across the board, regardless of the actual productive output of the worker. Inference has no such floor. The cost of a unit of AI labor is compute, energy, and a margin to the model provider — and all three are deflationary over time. When you remove the biological floor from the cost of cognitive work, margins expand structurally, and every company will be pressured by shareholders and basic capitalism to exploit that gap.

What’s important to mention is that this assumes a certain level of bullishness on AI progress. If you’re in the camp that AGI is coming, whatever that means for you, then you likely agree with this assessment. If you have a more conservative view on AI, it’s possible that this calculus won’t be done for a long time.

With that caveat aside, it’s clear that in this scenario, money will flow from firms needing inference to the model companies, namely OpenAI and Anthropic, and those companies will need compute, likely from Nvidia. Today, Nvidia trades at roughly 21x forward EPS, a multiple that assumes datacenter revenue growth decelerates meaningfully from here. But if inference demand follows the trajectory of token consumption, which has grown ~10x in a year, then current forward estimates are likely too conservative.

It’s generally accepted that GPUs have depreciation cycles of about 3-5 years, and Nvidia is also innovating on their products on a regular basis. If compute continues to ramp up, and we see mass adoption of inference as a replacement for labor, I fail to see how datacenter spend slows in the near term.

Subscribe now

Other (seemingly) Cheap Stocks in this Scenario

Micron, SK Hynix and Samsung are all glaring buys in the scenario that data center spend continues at this pace over the next 3+ years.

Micron ($MU) trades at 4x forward PE

SK Hynix at 4x forward PE

Samsung at 6.8x forward PE

“Markets are efficient bro..”

This is probably the thing that scares me the most being long a few of these names. The market is generally pretty efficient, and if it’s not moving on these forward estimates, it’s worth asking why rather than assuming everyone else is wrong.

For the memory names specifically, the market has real reasons to be cautious. Export controls on China remain a moving target, any tightening hits Samsung and SK Hynix directly. HBM is a cyclical product category with a history of brutal oversupply corrections, and the market has been burned before. There’s also a legitimate question about whether hyperscaler capex is sustainable at this pace, or whether we’re approaching a period where the Mag-7 pull back on spending as they digest existing capacity.

For Nvidia, the discount likely reflects a combination of those same capex sustainability fears plus the growing threat of custom silicon — Google’s TPUs, Amazon’s Trainium, and Microsoft’s Maia all represent real long-term share risk, even if Nvidia’s competitive position today is dominant.

I don’t think any of these risks are trivial. But I do think the market is pricing them as near-certainties rather than possibilities. If even half of the inference demand thesis plays out, these names are mispriced. All of this is to say that if you are a believer in AI, there are ways to play that and be exposed to that upside, but you should be watching what the market is telling you.

Disclaimer: This is not financial advice, do your own due diligence please.

Obviously there have been significant macro developments in the time since April of last year, namely the ongoing conflict with Iran that remains to be a murky situation.

Meta’s Legal Concerns

The main reason that has sold off so significantly is because of legal developments surrounding the mental impact on the younger users of their apps.(Reuters)

The prevalent bear case is that this is just the beginning and that there are 3000 similar cases looming in the California court system, let alone more globally. It’s also been posed as social media’s “Big Tobacco Moment.”

What I find to be the interesting question here is what the outcome of these court cases will be to the end product of Meta, attention. I think that while these cases, and the precedent set by the most recent case can obviously cost Meta billions in legal fees, the bigger question is how will all of this change Meta’s core business structure.

Will Meta be forced to make their platforms less addictive? Removing the endless scroll, curated feeds, and targeted advertising would certainly harm their business.

But if the age gate is moved up (currently 13yo in the US) to 16 or 18 years old, I don’t see this having a significant impact on their business for a few reasons.

1. Teens will find a way around age gates. As someone who grew up with video games, social media platforms, etc. it is easy to just lie about your age or sign up in your parents name.

2. I would expect that teenagers have far less disposable income, and therefore advertising to them, while valuable, is not that substantial of an income source compared to adults and seniors.

3. Teens have had social media for years now. A 13 year old today was born in 2013. They experienced the covid lockdown at 7 years old and are likely daily users of Instagram, YouTube, etc. If the age gate moves up to 16yo or 18yo, they will simply find ways around it, or just wait a few years and be back on social media as if nothing happened.

If the incoming regulation is a more structural change to their apps like what I mentioned previously (removal of the endless scroll, curated feeds, and targeted advertising) then I think this $150b+ market cap sell off is warranted to some extent.

However, I find this extremely unlikely. The endless scroll, curated feeds, and targeted ads have been the norm of social media since late 2019-early 2020. This level of sweeping regulation would have massive implications. Is it possible, sure, but far from likely.

What I find to be more likely is that META continues to sell off as the litigation continues and that this narrative is similar to Google’s antitrust concerns in 2023. Meta and Google will likely have to pay billions in legal fees, but this seems more like a short term headwind than a structural risk to their business.

Meta’s AI Spend Concerns

Another portion of the META sell off is likely due to their increased AI spending, as anyone who has followed Meta for some time now, the market is typically not a fan of Zuckerberg’s capex tendencies.

Meta joined other mag-7 companies in 2026, projecting massive capital expenditures on AI of $115-135b in FY26. This announcement was made in the Q4 2025 earnings report on January 28, 2026, and the stock is down ~30% since.

Meta is not alone in this sell off, as many of the mag seven stocks are down as well, but part of the reason why I think the sentiment/narrative surrounding Meta’s capex is so negative is because Zuck has shown a propensity to keep investing even when the market is telling him to stop. (see Reality Labs spending in 2021-2022.)

The flip side is that Zuck has also shown an inclination to allocate capital to shareholders via buybacks and dividends when the stock remains lower than historical levels.

Zuck has also shown a willingness to make the company leaner, with sweeping layoffs in recent years. Additionally, in the 2022 sell-off, Meta bought back roughly 168 million shares, roughly 6% of the outstanding shares. So, if we see an extended drawdown to lower multiples, (we’re already at ~17.7x FWD PE) I believe that they would get aggressive with buybacks again.

Meta’s Recent Stock Based Compensation Plan

Another key headline that got lost in the recent volatility is the SBC plan that Meta recently announced. It’s interesting to discuss because while it looked overly aggressive when it was announced, it looks even crazier after the company has dropped ~$150b in market cap.

The option plan’s strikes are extremely high, ~$2770 on average, which implies a 31% CAGR from current levels shows internal confidence on levels I haven’t seen before. Additionally, Meta’s CFO, Susan Li, is strongly incentivized to drive even more extreme returns for the company:

While a $7 trillion market cap seems hard to believe, the incentives are there, and you could have said a similar thing about NVDA going from $745 billion in ‘21 to $4.1 trillion today. Obviously it gets harder as the numbers get bigger, but I find it hard to be pessimistic about a company that dominates the social media space, and has ~1/2 of the world as part of its DAU base.

If management is this confident internally, the market's pessimism may be overdone.

Trade Recommendation and Position Sizing

What I learned from my first go-around with in 2022 is that the market can move against you far longer than you think it would. While I acquired ~100 shares for around $170-$180/share, and did very well on that investment, had I been more patient with my investment, I could have added to Meta sub $100, which I deeply regret missing out on.

With that being said, we don’t know if this will last nearly as long as the Reality Labs sell off. This could be a short blip in time where Meta falls ~30-35% off all time highs and recovers after the macro situation clears up. Or we could be in for a 50% drawdown and further war escalations could drive the whole market down.

That’s a long way of saying I have no clue how low Meta will go. The chart guys on Twitter seem to be confident it’s going to ~$475 or roughly another 10% lower from these levels. I have no expertise with charting, so I defer to their judgment, but I think that I’m going to start investing at these levels.

In terms of vehicle for this trade, I think shares are the only logical choice right now. The market seems to be far too volatile for LEAPS right now, and if you buy an ATM leap and META goes down even just 10%, your position is likely down close to 30% or more.

You can look at cash secured puts, and I have been, but you really have to tie a lot of capital up ~$50k per CSP, but the premiums are there. If you sell a January 2027 CSP, at $530/share, you get $7,200 in premium or roughly 13.5% on your money in that time period. If you get assigned, your effective cost/share is ~$458, which is not too bad, but obviously you tie up a lot of flexibility if the market continues to move against Meta.

Personally I will continue to buy shares as Meta goes down, but if there’s a significant narrative shift, and some of the legal headwinds subside, I will consider LEAPS, as I think Meta is positioned really well amongst its peers in terms of growth prospects vs. current valuation multiple. Currently, I would start with 1/4 of what a full position looks like, and have cash ready to deploy if Meta continues to fall.

If you enjoyed this writeup, consider subscribing, as it’s totally free and helps me out a ton. I post about stocks, prediction markets, and finance in general.

Disclaimer: This is not financial advice, do your own due diligence please.

Until now. Fundrise, a company that began by democratizing access to large scale real estate investments, offered access to two of the premier AI labs, OpenAI and Anthropic, via their ‘Innovation Fund’ with the ticker

While the fund offers exposure to many other high profile private companies, the main excitement from retail investors has been the stakes in Anthropic and OpenAI. Below are the top 10 holdings by weight in the fund.

With both OpenAI and Anthropic at the top of the list, it’s easy to see why has been bid up by retail investors. The problem is the level to which it’s been bid up.

The Net Asset Value (NAV) of this portfolio is $18.97/share, and retail investors have bid the fund up to $270/share as of writing this on Tuesday. This means the fund is currently trading over 14x the NAV of the assets it is meant to track.

The reason why? Well there is undeniably high retail demand for these AI companies, but more importantly for this specific phenomenon is that there is a massive supply/demand imbalance going on right now with VCX specifically.

The supply/demand issue is happening because:

“The majority of the fund’s shares remain subject to a lockup agreement. For the fund’s roughly 100,000 investors who bought prior to the market debut, VCX shares acquired before Feb. 20 are restricted from being sold for six months after the listing, the filings show.” (Investing.com)

This means that the float, or total shares being traded are extremely low, which means that this sort of ticker can be extremely volatile. As I’m writing this, it’s down to $258 from $270 just minutes ago.

Now obviously you don’t want to be invested in Anthropic/OpenAI through VCX right now as it’s just trading at such a premium to their NAV and it’s going to be highly volatile for the near term.

So, what can you do?

Well you can keep waiting, as Anthropic and OpenAI are expected to IPO this year, but obviously that may not happen depending on a myriad of factors.

What I am doing, and what you can do as well is have positions in large companies like Google, Amazon, and Microsoft, all of which are off the 52wk highs by a decent margin.

But if you wanted to bet on Anthropic specifically, and in a big way, I think Zoom ($ZM) is a good way to play it. They invested $51m in May of 2023 when Anthropic was valued at ~$4.5b. Since then, Baird estimates Zoom's stake is now worth $2–4 billion depending on dilution assumptions, against Anthropic's $380B valuation.

With all of that in consideration, Zoom’s stake in Anthropic is likely worth ~15-20% of their entire EV ($3.3b vs. $17b), meaning that they offer the most direct way to invest in Anthropic as it stands.

Obviously this means you will have to be exposed to Zoom in your portfolio, but I think this ownership stake will continue to dwarf their own business to the point where the stock will move significantly on news concerning Anthropic’s products, fundraising, and timeline for their IPO.

For direct ownership in OpenAI I think is the clearest as while its only ~5% of MSFT’s EV, they have ~27% ownership in OpenAI, so there’s significant upside once OpenAI IPOs.

If you enjoyed this writeup, consider subscribing, as it’s totally free and helps me out a ton. I post about stocks, prediction markets, and finance in general.

This is my concern with Adobe. Is it a dying software company, made entirely irrelevant by Generative AI & image/video creation? Or is the narrative oversold and are they going to turn this ship around?

Buying (Paypal) a few years ago was a prime example of me buying a stock because it was trading at a low multiple compared to peers and on a historical basis. What did $PYPL do since it traded at this great discount?

Well since late ‘22, where it was more than 75% off of its peak, it has gone down a further 35%, and is trading at such a low market cap that it is being viewed as an acquisition target.

All of this is to say: If you bought Paypal nearly 80% off of its peak in late 2022, you’re not very happy right now. You’d be down ~35% in a time where the S&P 500 returned over 70%.

My point is that buying out of favor stocks, or ‘turnaround’ stories is smart, but only if that turnaround is actualized, not just projected.

This brings me to Adobe today. I think they are in a similar position to was in late 2022.

The stock has cratered over 60%, weighed down by moat erosion fears and the looming question of whether generative AI renders Adobe obsolete.

Is This Narrative Right?

The obvious next question is whether or not this narrative is right or if its more AI-eccentrics pushing narratives that warrant their own lofty valuations.

Taking their side for a moment, if I were to argue that Adobe is set to be disrupted by generative AI image and video generation, my main arguments would be:

• The barrier to entry for "good enough" creative work has collapsed. Midjourney, Nano-Banana , Canva's AI suite, and others can now produce output that would have required a trained Photoshop user three years ago — hitting the Business Professionals & Consumers segment hardest.

• Where does good enough settle? If 80% of creative work gets done by AI tools that don't require Adobe, and only the top 20% of professional work remains in the ecosystem, the TAM for premium creative software shrinks

• AI-native tools accelerate churn in the growth segment. The users Adobe needs to acquire to expand TAM are the same users most likely to find standalone AI tools sufficient.

If I were to argue against this narrative, my main points would be:

• The consumer has shown a propensity to be anti-AI (see the Coca Cola ads in 2024-2025) and various social media accounts for NBA/NFL teams. This is something I expect to continue — and I expect that in the future, AI media will have to be watermarked or otherwise disclosed so that the consumer knows if what they are watching is real. See the announcement and subsequent response below:

• Video editing — not video creation, is a difficult process that AI cannot simply one-shot. Much of content creation is taking longer footage and turning it into smaller pieces of content that keep the viewers attention over time. Additionally, there’s a lot of work done before video even gets to the editing stage. Blocking, camera placement, lighting, are all things that will be difficult, if not impossible, to recreate via prompt.

• Adobe is not just Photoshop — a large part of Adobe’s negative sentiment is the idea that these image generation tools will remove the need for Photoshop, and therefore much of Adobe’s subscriber base will churn. I think this is far from the truth for many reasons. Namely, there have been high quality photoshop clones for years: Canva, Photopea, Affinity, etc. If Adobe’s revenues were as Photoshop-centric as this narrative makes it seem — we would see impacts on revenue way before Nano Banana and Midjourney.

• Additionally, Adobe has made some progress towards including AI tools within their software. Firefly is Adobe's family of generative AI models built directly into Creative Cloud products, offering text-to-image generation, generative fill, and generative expand within existing workflows.

• Adobe's key differentiator is that Firefly was trained on licensed Adobe Stock content, making it "commercially safe" for enterprises, and they've even offered indemnification to enterprise customers against IP claims, something no other major AI generator can match.

Now that I’ve given you both sides, you have to decide what scenario you believe is most likely over the next decade. Will the next CEO be more of the same — stagnancy, more moat erosion? Or will they usher in a new era for Adobe, fueled by acquisitions of leading AI tools, combined with their commercially safe licensed content?

I think the next few dominos to fall will be very deciding for the future of Adobe. I am hoping they an external candidate for the CEO position that has a clear vision for how Adobe can position themselves against these new generative tools.

I am also hopeful that they can make acquisitions of certain AI tools, like Midjourney, that can strengthen their value proposition to consumers such that they can be a one-stop-shop for enterprises, entrepreneurs and creatives alike.

Scenario Analysis: Pricing in the Tail Downside Risk

Tail Downside: 5% Probability

Given that there is a scenario in which the consumer is unable to differentiate between AI generated content and human created content, it does make sense to price in , even if I view it as extremely unlikely.

Given that devastation that would be to Adobe as a company, I think an almost total wipeout of the current market cap. (-94%)

I assign this a 5% probability because it requires simultaneous breakthroughs across multiple AI domains — video editing, sound design, motion graphics, enterprise marketing orchestration — that are each individually far from being solved.

Bear Case: 20% Probability

AI cannibalization accelerates faster than expected. Creative Cloud growth decelerates to low single digits as Canva, Figma, and free AI tools chip away at the user base. The CEO transition creates extended uncertainty — the board takes 12+ months to find a successor and the interim period is marked by strategic drift. Enterprise customers delay renewals as they evaluate AI-native alternatives. Adobe Stock revenue collapses further. Multiple compresses to 8-9x forward earnings, reflecting the market’s view that Adobe is a melting ice cube.

Base Case: 50% Probability

Adobe continues growing revenue at 10-12% annually, with AI-enhanced subscriptions (Firefly, GenStudio, higher Creative Cloud tiers) more than offsetting the decline in legacy stock assets. The new CEO articulates a clear AI strategy that the market can underwrite. Margins remain stable at 46-47%. The multiple re-rates from ~10x to ~20x forward earnings, which is still well below historical averages but reflects renewed confidence in durability. At $23.50 in forward EPS × 20x = $470.

Bull Case: 25% Probability

New CEO drives a sharpened AI-first strategy that clearly differentiates Adobe from free tools. Firefly becomes the enterprise standard for commercially safe, IP-clean AI content generation. Video editing workflows integrate AI assistance (auto-cuts, AI-driven color grading, intelligent audio mixing) that makes Premiere Pro indispensable rather than replaceable. Operating margins expand to 48-50% through cost optimization. Revenue accelerates to 14-16% growth. Multiple re-rates to 28-30x forward earnings, still below the 10-year average of 40x. At $25+ EPS × 28-30x = $700-750.

Trade Recommendation + Catalysts to Watch

Read more

Warner Bros. Discovery: Recent Context

Before discussing Netflix as an investment opportunity, it is worth establishing the background of the company it is in the process of acquiring, Warner Bros. Discovery (WBD).

WBD was formed through an April 2022 merger between WarnerMedia (HBO, HBO Max, Warner Bros. Pictures, DC Entertainment, CNN, TBS, TNT, and Cartoon Network) and Discovery (Discovery Channel, HGTV, Food Network, TLC, Animal Planet, and Discovery+).

In June 2025, WBD announced plans to separate into two independent entities: a Streaming & Studios division (Warner Bros., HBO, DC Studios) and a Global Linear Networks division (CNN, Discovery, and related assets).

By December 2025, WBD had entered into an agreement to sell its Streaming & Studios division to Netflix for $82 billion. Paramount Skydance made multiple counter-bid attempts(for the entire company), and as of February 11th has raised its tender offer to $108 billion, including a commitment to cover the breakup fee should WBD walk away from the Netflix deal.

The distinction worth highlighting is that Netflix is selectively acquiring the Streaming & Studios division, the higher-value assets, while Paramount’s bid encompasses the entire company, including the Global Linear Networks segment.

In my view, Paramount’s offer appears less strategically motivated and more reflective of a need to maintain competitive relevance in the streaming landscape. Paramount has also attempted to frame the Netflix acquisition as an antitrust concern, positioning its own bid as a more regulator-friendly alternative by pledging to keep WBD’s entities operationally separate.

From Netflix’s perspective, the real value lies in the intellectual property housed within the Streaming & Studios division — franchises like DC Comics, Game of Thrones, and Dune. Netflix appears to have a clearer strategic rationale for how these assets integrate into its business, and that clarity of vision is a meaningful part of why I find the company compelling as an investment, particularly following a significant downward re-rating.

Investment Thesis: Growing their Audience & Increasing Customer Retention

Netflix’s core business has been steadily growing over the years. Their revenue in the fiscal year ending December 31, 2021 was $29.7 billion, and over the past four years it has grown to over $45 billion, at a CAGR of 11%.

Looking to the future, Netflix has decided to invest in acquiring intellectual property via the WBD acquisition that they really haven’t had in the past outside of the hit show, Stranger Things. I think this makes sense, as the problem with subscription based companies like Netflix, is that they will lose a lot of users to churn when shows like Stranger Things stop airing.

By acquiring WBD and its intellectual property, which includes hit shows like Game of Thrones, Netflix protects itself from churn while growing the audience of people who would want a subscription to their favorite shows and movies.

They protect the user base that they already have, while growing it substantially, and they also gain the ability to package HBO Max and Netflix at a cheaper price to consumers, similar to the Disney+/Hulu business model.

I believe that Netflix truly knows how to monetize streaming, as they have multiple tiers of subscriptions with and without ads, and have largely been able to crack down on password-sharing. If they can continue to dominate popular culture and acquire fan favorite shows and movies, they can increase their pricing power as well as the stickiness of each subscriber.

In addition to this, I believe there is a tail outcome in which movies quickly go from theaters to in-home streaming, and that Netflix is going to be a major driver of this. The CEO has already said that he wants movies to go from theaters to streaming quicker, claiming that it’s not ‘consumer-friendly’ to have long exclusive windows.

If Netflix can actually make this happen, especially with DC Studios as part of their library, I think this would drive a lot more revenue to them. From a unit economics perspective, the movie theater experience (~$20/ticket, family of four so ~$80 in tickets + $15-25 in terms of snacks/candy) is much more expensive than the $20-25/month streaming costs. If movies hit streaming within a few weeks of being released I think Netflix could really take market share from in-person moviegoers.

DC Studios could also be a major driver of revenue for Netflix, as we saw the MCU was a major financial boon for Disney in the late 2010s, and the DC’s version of that is just ramping up since the initial Superman movie released in mid-2025. I think that DC Studios is being run very well by James Gunn, and it's a major positive for the future of DC Studios that he was such a large part behind the success of Marvel’s Cinematic Universe in the late 2010s.

Back of the Envelope Math: What’s the Upside?

Just for fun, I thought I’d do some back of the envelope math to see what Netflix’s earnings/valuation could look like in certain scenarios.

In terms of my assumptions, I believe they are defensible and I will try to explain each one in detail here.

Churn Rate (WBD Overlap) refers to the amount of WBD users who already have a Netflix sub. This figure is from Netflix themselves, and is part of their legal argument to regulators. More below:

Adjusted Sub Count is taken by adding Netflix subs (325m) with WBD subs (128m) and subtracting the amount of WBD subs who would remove a sub because they already pay for Netflix. This is done to avoid double-counting.

Annual Rev/Customer is where I think this trade comes to life. If you average out the annual rev/customer from Netflix ($134) and WBD ($85) you get ~$110. I put this as the bear case to be conservative, but I think it’s highly unlikely that Netflix doesn’t raise prices if this deal goes through.

I think an average revenue per user of $160/year or $13.33/month is entirely reasonable, especially considering the increase in content provided with that subscription. This would be a 19% increase from Netflix’s current annual revenue per user.

In the bull case, I put $210 as the ARPU, as this would be a massive increase in subscription cost, but Netflix may have the pricing power to pull this off.

Considering that 80% of WBD subscribers feel that they can afford both the $134/yr for Netflix and the $85/yr for WBD, totaling $219/yr, maybe paying $210 isn’t that crazy?

Net Profit Margin I assumed to be stable because Netflix has operated in the streaming business for years and I don’t expect that this acquisition will materially affect costs.

NTM PE Ratio I believe will increase from 25 to 30 in the case that this acquisition goes through, and if Netflix can successfully monetize their combined product of Netflix and WBD, I think it could even re-rate to 35+.

I think based on these assumptions, Netflix provides an asymmetric opportunity — -25% downside with upside to ~100% in returns over the next few years.

It’s important to note that these scenarios all assume the acquisition goes through successfully which is not a given. They also do not take into account the benefit of natural subscriber growth that is expected of these companies.

Near Term Catalysts

While I don’t necessarily disagree with the antitrust concerns, if this deal does pass, I think Netflix will definitely have an even more dominant position within the streaming space. With IP like Dune, Harry Potter, Game of Thrones, and DC Studios, and Netflix’s belief that the gap between a movie releasing in theaters and going to streaming will close, I think that we’re going to see even more people just skip the theater and decide to wait till the movie hits streaming. See co-CEO Ted Sarandos’ comments below.

It is important to note that while I believe the acquisition will go through, traders on Polymarket have the odds at only ~50% that Netflix will be the company to acquire Warner Bros by mid-2027.

This acquisition and any legal developments on it are definitely going to be catalysts for Netflix as a company. While I like this acquisition a large part of my investment thesis, I do not believe there is significant downside if it falls through given Netflix’s large pre-existing business, especially from this slightly depressed valuation.

The shareholder vote on the Netflix/Paramount acquisition debacle is likely going to be in mid-late March, so I would expect clarity on the situation by at least April, but the legal process could take much longer. Especially considering David Ellison (CEO, Paramount), the son of Larry Ellison (CTO, Oracle), are lobbying to Trump that the acquisition by Netflix should be blocked, citing the aforementioned antitrust concerns.

Costs

From 2021 to 2025, operating margins have gone from 20.8% to 29.5% respectively, signifying that has been able to keep costs under control while ramping up revenues. Given that their levers to pull in terms of revenue drivers are increasing subscription costs + event-style content, this makes sense intuitively.

I think in terms of costs you should be watching, it’s really just cost of acquiring new content. While Disney/Hulu+ is a competitor to Netflix, I think their main competitor is YouTube.

The reason YouTube stands out to me is that YouTube’s content is primarily user-generated, meaning that they are not undertaking any costs besides the service of hosting all of this content. This means that their job is just serving the right content to the right consumer, and keeping that consumer on platform for as long as possible.

So, in that same vein, as investors in Netflix, the cost you really want to watch are the costs of acquiring new content, which this acquisition is doing. Does acquiring and getting content like Game of Thrones, DC Studios, Harry Potter, etc. drive new users, and maintain them over time? My answer to that question is yes.

With the release of HBO’s newest Game of Thrones spinoff, A Knight of The Seven Kingdoms, my Twitter, YouTube shorts, Instagram Reels, etc. timelines have been full of content about the show, with millions of impressions on each post. Game of Thrones’ other spinoff, House of the Dragon also is expected to release mid-2026.

Additionally, DC Studios has Lanterns coming out in the summer of 2026, and HBO Max also has a Harry Potter show coming out in early 2027.

With all of these different IPs releasing year round, you have to think subscribers will opt to pay for the annual subscription, saving slightly on subscription fees, but reducing the churn that naturally comes with streaming services.

The acquisition of WBD and these assets makes sense because it truly offers Netflix to have a full calendar of content releasing, which would materially reduce both churn and the seasonality of revenues around shows like Stranger Things.

Costs and Debt of Acquiring Warner Bros. Discovery Streaming Division

A key concern is the debt load Netflix is taking on with this acquisition. Netflix is taking on $42b bridge financing for this deal, and has paused share repurchases.

This is likely a large part of their re-rating downwards, but I think investors are missing the forest for the trees here. Companies make investments in the expectation that future revenues will pay for them multiple times over. I believe that Netflix’s investments into this differentiated IP and content library is worthwhile and I do think it’ll pay off in the years to come.

Valuation

has sold off quite a bit, down over 40% from its 52wk high of $134.12. It’s now trading at a 3 year low PE ratio of 24x NTM earnings. I think the reason for the sell off is primarily concern over the acquisition of and the uncertainty regarding that. You can see that diverges from the broader market in early December of 2025, right when the acquisition was announced.

I think this fear is pretty overblown, as even if the deal does fall through, Netflix has shown that they understand how to lower churn and maintain subscribers, as they have ramped up the number of event-like showings over the past 12-18 months.

Since January 2025, Netflix has a WWE Residency, airing WWE Raw live every Monday night. Netflix has partnered with the NFL to air games exclusively, as well as the MLB, where they will cover opening night this coming March. They also broadcasted Alex Honnold free climbing Taipei 101 in Taiwan just a few weeks ago. I think Netflix will continue this trend, so that they have more sports/YouTube-like content on platform, and attract viewers of that style of content.

If the deal goes through, I don’t see how Netflix doesn’t immediately re-rate upwards to ~30x NTM PE. This alone would drive returns of about 20%, not taking into consideration the new subscriber base acquired via WBD.

Positives

1. Substantial user-base regardless of acquisition fall through.

2. Increased events and integration of sports

3. Lowering movie theater exclusivity periods

Negatives

1. Regulatory concern with acquistion

2. Platforms like YouTube/Twitch/Instagram Reels eat into watchtime for younger generations — meaning there is less of an appetite for longer form content on Netflix. If the trend of younger generations being less into TV shows/movies continues, this could be a long term risk for companies like Netflix and Disney.

What would an extreme bear case look like?

As part of my back of the envelope math, I failed to show what the downside would be if the deal fell through entirely, so I thought I’d add this section.

In this extreme bear case, share price falls from $76.87 to $51.57, a 32.91% loss. This is driven by no subscriber growth, acquisition not going through, and average revenue per user staying flat. I think the likelihood of this scenario is exceedingly low, but I thought it could be worthwhile just to show what that might look like.

The Trade:

I personally think that Netflix is a quality company that will continue to grow in subscriber count over time, and that this acquisition is just a way for them to speed up that process while gaining content and IP that will increase customer retention over time. Their continued development and iteration of event-style content will enable them to compete with the likes of ESPN, YouTube, and Twitch in the years to come.

Read more

Anchoring, specifically price-anchoring, is a cognitive bias where individuals rely too heavily on an initial piece of information (stock price) when making decisions.

We’ve all done it. “ has gone up over 300%! It can’t keep going up.” Yet it did. From October 9, 2022 to November 5, 2023, Nvidia went up over 330%. Since then, it’s gone up another 273%.

The conservative thing to say was that it’s run up too much.

That’s pretty much how I thought about it at the time. It’s safe to say I am regretting that thought process.

But I think the stock market has provided an opportunity to right that wrong with two stocks, SK Hynix and Samsung. If you follow markets, you’ve likely seen or heard about SK Hynix’s performance over the last year. SK Hynix is up 375% in the past 12 months, and up 251% over the past 6 months alone.

The Story

SK Hynix is the dominant supplier of HBM, (High Bandwidth Memory) controlling over 60% of the market. They are the primary supplier for Nvidia’s Blackwell and Rubin platforms.

In addition to HBM, SK Hynix produces DRAM and NAND Flash, both of which’s demand has increased significantly with the AI boom. DRAM contract prices are up roughly 170% year over year, and retail RAM kits (16-32gb) are up 240% over the past year.

Recently, even DDR4 and DDR3 ram prices have even surged as AI demand is actually extending into older DRAM models. SK Hynix has had to lower production of consumer DRAM to focus on AI memory, which is furthering this supply shock.

NAND Flash memory, which is behind the SSDs required for consumer PCs as well as data centers, have increased as well. A 1 terabyte NVMe drive that cost $65 in early 2025 is now retailing for over $130-$150.

SK Hynix, who develop and manufacture all of these products announced that their entire DRAM and NAND production capacity for 2026 was already sold out.

The CEO of Intel, Lip-Bu Tan, said today at the Cisco AI Summit that memory is the biggest constraint for AI right now, and that there is no relief til 2028. If someone on the inside of the AI boom is saying this, I think this makes a compelling argument for both SK Hynix and Samsung’s continued outperformance as the two clear leaders in the HBM and DRAM market.

While hyperscalers like Nvidia, Google, Amazon, are drivers of DRAM demand, AI video-generation of longer form content will ramp up demand for NAND flash storage that powers SSDs.

The typical Sora or Meta AI video is currently only 15-30 seconds long. Think about when those videos become 15 minutes to 1 hour in length, and they are being generated every second.

The Numbers

Morgan Stanley estimates Samsung’s 2027 operating profit to be $218.5 billion, which would make them the most profitable company in the world. Samsung’s current market cap is $771.67 billion, meaning that they are trading at roughly 3.5x 2027e operating profit. Quite cheap if you ask me!

Morgan Stanley estimates SK Hynix’s 2027 operating profit to be $155 billion, which would rank them with the likes of Mag-7 darlings Google, Meta, and Apple, while their market cap is only $455 billion. (2.9x 2027e operating profit)

If we assume that these figures from Morgan Stanley are accurate, and apply a 10-15x multiple to both Samsung and SK Hynix, they both provide attractive returns at their current levels.

For these reasons, I believe SK Hynix and Samsung will be the next two (non-IPO) companies to join the trillion-dollar market cap club, and that they both provide exceptional opportunities for returns over the next 2-3 years.

The Trade

The difficulty in this trade is actually not in the thesis or financial projections, I actually think it is in execution. Currently there are no US-based listings for Samsung or SK Hynix. This means that in order to gain exposure to these two companies, your options are somewhat limited.

Read more