The AI Bubble Question

Wall Street hijacked the stock market 200 years ago. 

Now in 2026, they're coming for YOUR digital assets.

Bitcoin was supposed to be peer-to-peer. No banks. No middlemen.

Not anymore.

BlackRock owns more Bitcoin than most countries. 

Fidelity's ETF hit $10 billion. 

JPMorgan called Bitcoin a "fraud" — now they run billions in tokenized assets. 

They ARE crypto now.

Every time you hit "Buy" on Coinbase, you're trading at their prices that they've already positioned themselves for the biggest returns. You're fighting over scraps.

It's the 2008 playbook. 

Wall Street sold mortgage-backed securities to retail, then shorted them and made billions while people lost their homes.

But there's a way to operate outside their system.

Tan Gera, ex-Wall Street banker and CFA Charterholder, walked away after discovering their two-tier system. 

Now, his 35-person research team helps 3,000+ investors access opportunities before Wall Street marks them up 100x.

Click here to watch the free presentation now →

For educational purposes only. Results will vary. DM Intelligence LLC is not liable for losses.  

Today is Jerome Powell's last day as Chair of the Federal Reserve. By Saturday, Kevin Warsh holds the gavel. And in the middle of that historic transition, Michael Burry, the investor made famous by The Big Short for calling the 2008 housing collapse, is publicly comparing the current market to the final months of the dot com bubble.

His evidence. The Philadelphia Semiconductor Index is up over 65 percent year to date. Up more than 10 percent this week alone. Stocks rising, in his words, on a two letter thesis that everyone thinks they understand. He is not predicting an imminent crash. He is observing that the same psychology that defined late 1999 and early 2000 is visible in the current price action.

Meanwhile, JPMorgan upgraded its S&P 500 target to suggest the index could hit 8,000 by year end. That would be roughly 8 percent above Friday's record close of 7,398. Their thesis rests on continued earnings growth, geopolitical de-escalation, and the AI capex cycle continuing to drive infrastructure earnings higher.

Both calls cannot be fully right. But here is what most investors miss. Both can be partially right. The market can grind higher for another six to twelve months on continued AI capex flows, and then experience a meaningful repricing when the cycle peaks. Bubbles do not end at the moment they become identifiable. They end when the marginal capital that was sustaining them runs out, and that timing is notoriously difficult to forecast.

Your job is not to call the top. Your job is to construct a portfolio that participates if the rally extends and survives if the reckoning arrives. Today I want to walk you through the genuine evidence on both sides, the institutional framework for positioning into a late stage rally, and the specific moves that protect you regardless of which scenario plays out.

THE EVIDENCE FOR THE BUBBLE THESIS

Let me start with what Burry and other skeptics are actually pointing to, because the case is more substantive than the financial media coverage suggests.

First, the breadth of the rally is narrower than the headlines imply. Year to date through early May, the S&P 500 is up roughly 5.7 percent. Of that gain, the top 10 stocks contributed 5.1 percentage points. The top 20 contributed 6.5 percentage points. That second number is actually larger than the index return, which means the bottom 480 stocks in the S&P 500 are net detractors from index performance this year. Alphabet alone has contributed 1.27 percentage points to the index return. That is more than 20 percent of the year to date gain from a single stock.

This is the textbook definition of a narrow rally. When 10 stocks deliver more return than the other 490 combined, the index is functionally trading like a concentrated portfolio. The diversification that retail investors think they own when they buy a broad market index fund has been mathematically eroded.

Second, the valuation multiples have expanded faster than the underlying earnings. The forward 12 month price to earnings ratio for the S&P 500 stands at 20.9. That is above the five year average of 19.9 and above the ten year average of 18.9. More importantly, it is materially higher than the 19.7 level recorded just six weeks ago at the end of March.

Multiple expansion in a rising market is normal. But multiple expansion accelerating in the late stages of a cyclical advance is a classic late cycle signal. Investors are paying more per dollar of earnings even as the absolute earnings level keeps rising. That requires either continued strong earnings growth to validate the higher multiple, or a willingness to pay even more in the future.

Third, the capex commitments from the AI infrastructure leaders have reached the kind of magnitude that historically marks cycle peaks. Meta's Q1 print included guidance for 2026 capital expenditure of 125 billion to 145 billion dollars. CoreWeave pulled up its 2026 capex range to 31 to 35 billion dollars from 30 to 35 billion dollars. The hyperscalers, in aggregate, are deploying capital at a pace that would have been considered reckless just three years ago.

The bull case for that capex says the demand for AI compute is real, the returns on the capital will be excellent, and the companies deploying it will be the dominant infrastructure providers of the next decade. The bear case says we have seen this movie before. Telecom companies in 1999 deployed capital at unprecedented levels building fiber optic networks for an internet that did not yet have the demand to fill them. The fiber was eventually used. The companies that deployed it mostly went bankrupt first.

Fourth, the speculative behavior at the margins is starting to look familiar. Intel stock more than doubled in April. Fluence Energy gapped up almost 30 percent in a single session on an earnings beat. SpaceX is heading toward an IPO that has the entire retail trading community salivating. These are not necessarily wrong calls. They are signals that the appetite for risk has expanded faster than the underlying earnings have grown.

THE EVIDENCE FOR THE CONTINUATION THESIS

The bull case is also substantive, and intellectually honest investors need to engage with it before dismissing the rally as obvious euphoria.

First, the earnings growth is real and broad based. Q1 2026 earnings growth ran at roughly 15 percent year over year. The forward expectations for Q2 through Q4 are above 20 percent. Even if those forward numbers prove optimistic by a few percentage points, the absolute earnings level continues to grow at a healthy clip. A market priced at 20.9 forward earnings with 15 percent earnings growth is not nosebleed expensive. It is fully valued, which is different.

Second, the AI capex cycle is genuinely transformative in a way that the late stage telecom build out was not. The applications running on top of the infrastructure are generating measurable productivity improvements. Enterprise spending on AI tools and platforms is accelerating. Consumer adoption of AI products has compressed timelines that historically took years into months. The earnings flowing to the infrastructure layer reflect actual end demand, not speculative buildout for buildout's sake.

Third, the macro setup, while complicated, is not signaling imminent recession. Q1 2026 real GDP grew at a 2.0 percent annualized rate. The labor market added 115,000 nonfarm payroll jobs in April. Unemployment sits at 4.3 percent. Corporate balance sheets are strong. Consumer credit metrics are stretched but not breaking. None of these data points suggests the kind of macro inflection that would force a sudden equity revaluation.

Fourth, the rate environment is friendlier than it has been at any point in the last 30 months. The Fed funds rate sits at 3.50 to 3.75 percent, down from a peak above 5.25 percent. Even without further cuts, the policy rate is no longer actively restrictive. Equity multiples typically expand when the policy rate falls below the perceived neutral rate, which most economists place around 3 percent. We are close enough to that level that the discount rate effect on equity valuations has flipped from headwind to mild tailwind.

THE FRAMEWORK FOR LATE STAGE POSITIONING

Here is how institutional desks position into a market that has all of these characteristics simultaneously. They do not bet on the timing. They build portfolios that perform adequately across a range of scenarios and they manage exposure size, not direction.

Step one. Define your scenario set. For the rest of 2026, the plausible outcomes are roughly as follows. Scenario A, the rally extends into year end and the index reaches the JPMorgan target near 8,000. Scenario B, the rally pauses and the index trades sideways through the second half. Scenario C, a meaningful correction of 15 to 20 percent occurs in the back half of the year as the AI capex narrative confronts execution challenges or as the new Fed Chair triggers a regime shift in rate expectations. Scenario D, a more severe drawdown of 25 percent or more occurs if multiple risk factors compound.

Each of these scenarios has a non trivial probability. None is greater than 40 percent in my estimation. The point is not to assign precise odds. The point is to recognize that all four are realistic and to build a portfolio that does not catastrophically underperform in any of them.

Step two. Size your exposure to your tolerance for the worst plausible outcome, not your expectation of the most likely one. This is the single most important risk management discipline I learned on the trading desk and the one that retail investors most consistently get wrong. Position sizing is determined by what happens if you are wrong, not by what you think will happen if you are right.

Practically, this means asking yourself what the index doing minus 25 percent in the next nine months would do to your total portfolio, and whether you would be able to ride it out without panic selling. If the answer is no, your equity exposure is too large regardless of what you think the most likely outcome is.

Step three. Implement a barbell rather than a single point bet. The barbell holds high quality, lower volatility exposure on one end and concentrated tactical exposure on the other end, with very little in the broad middle. The high quality end captures the macro tailwinds without being concentrated in the late stage AI rally. The tactical end captures the asymmetric upside if the rally extends, but it is sized small enough that a 30 to 50 percent drawdown in those positions does not damage your overall plan.

A practical implementation looks like this. Sixty to seventy percent of your equity allocation in quality compounders, dividend growers, and broad international exposure. Twenty to thirty percent in your AI infrastructure tactical sleeve. Ten percent in a defensive sleeve of gold, short term Treasuries, and managed futures, which historically perform well during equity drawdowns.

Step four. Use cash strategically. The 10 to 15 percent cash allocation that I have been emphasizing in recent issues serves a specific purpose in this environment. It is not parked there because cash yields well, although a 4 percent T bill is genuinely attractive on a risk adjusted basis. It is parked there because cash is optionality. If the JPMorgan scenario plays out and we reach 8,000, you will have left some upside on the table but you will be fine. If the Burry scenario plays out and we see a 25 percent drawdown, that cash allocation will be the most valuable asset in your portfolio. You will be able to deploy it at much better prices into a market that has just punished overleveraged investors.

THE ACTION ITEMS

Tonight, before the weekend, take 30 minutes and run this audit on your own portfolio.

Pull every account into a single view. Use a tool like Empower to aggregate your holdings across brokerages, retirement accounts, and any other invested capital. The aggregated view almost always reveals concentration that is not visible account by account.

Calculate the percentage of your equity exposure that is direct AI infrastructure related. If you cannot answer that question quickly, the position is too opaque to manage.

Identify what your portfolio looks like under each of the four scenarios above. You do not need precise figures. Round numbers are fine. The exercise is to feel the range of outcomes, not to model them precisely.

Decide whether your current allocation is consistent with your tolerance for the scenario C and D outcomes. If you would not be comfortable holding through those outcomes, trim now. The trim does not need to be dramatic. Even reducing your AI infrastructure exposure by 25 percent and parking the proceeds in short duration Treasuries materially improves your scenario C and D outcomes.

If you use M1 Finance to manage your portfolio, set up a separate tactical pie for your AI exposure with a defined target percentage. The discipline of having a target percentage forces you to trim when the position outperforms its target weight, which is exactly the behavior that protects you in late stage rallies.

ONE LAST THOUGHT

Powell hands the gavel to Warsh tomorrow. Burry is comparing the market to 1999. JPMorgan is targeting 8,000. The earnings season just delivered an 84 percent beat rate. Oil is volatile, the Middle East is uneasy, and the AI capex cycle continues to deploy capital at a pace that strains historical comparisons.

This is the environment. None of these factors will resolve in a clean, identifiable moment. The new Fed Chair will not give you a clear signal on day one. The bubble, if it is one, will not pop on a particular date. The earnings growth will not break in a single quarter.

Your edge is not in predicting the resolution. Your edge is in building a portfolio that does not require a particular resolution to perform adequately. That is what the institutional framework gives you. It trades the false comfort of conviction for the real protection of optionality.

Most retail investors will spend the next six months trying to call the top, or trying to ride the rally as long as possible, or oscillating between the two. The investors who do well across a range of outcomes will be the ones who set their position sizes today based on what they could afford to lose, and then went and did something else with their weekend.

Reply with the keyword AI and I will send you the AI Exposure Audit, a one page worksheet that walks through every line item above with specific position size targets and the trigger points that would shift you to more defensive positioning. It is the same framework I am using with private clients this week as Powell hands the gavel.

Until Sunday, stay strategic.

Taylor Voss

Money Systems Lab

ChatGPT is a superpower if you know how to use it correctly.

Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.

Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.

Download the free guide.