In today’s Finshots, we explain why the famous short seller is going short on AI when everyone is going bullish on AI stocks.

But before we begin, a quick reality check. Insurance matters only once in your life: the day your family needs to make a claim. And that’s usually when people realise they bought the wrong policy which don't fit their needs, missed a crucial detail, or trusted an agent who didn’t tell them the full story. But you shouldn’t have to learn this the hard way!

So starting tomorrow, we’re hosting a free 2-day Insurance Masterclass built around real claim cases we’ve handled and simple, practical rules for shortlisting the right plan.

📅 Tomorrow (Saturday), 29 Nov ⏰ 11 AM: Life Insurance
How to protect your family, choose the right cover amount, and understand what truly matters during a life claim.
📅 Sunday, 30 Nov ⏰ 11 AM: Health Insurance
How hospitals process claims, common deductions, the mistakes buyers usually make, and how to choose a policy that won’t disappoint you when you need it most.

If you want clarity, real-world insights, and simple steps you can apply immediately, you shouldn’t miss this one!

👉🏽 Click here to register, seats are filling fast.

The Story

Almost three weeks ago, Michael Burry, the short seller who bet against the housing market before the 2008 crisis, closed down his asset management company — Scion Asset Management. He deregistered the hedge fund on November 10, 2025, which officially meant he’d be stepping away from professional money management. When the company was closed, the news couldn’t help but speculate if the scion of Scion was actually going to retire, which is why he came out clean stating that he’s clearly not going to.

That led to the next guessing game: Why did he step away and what comes next?

It started with the new rules by the Securities and Exchange Commission (SEC or America’s version of SEBI) — a tougher regulatory climate for registered investment advisors (RIAs) between 2023 and 2025. Things like mandatory quarterly fee reports, audited statements, a heavier paper trail of positions and higher transparency, combined with the fact that Michael Burry’s name packs a punch, meant he couldn’t do what hedge funds existed to do in the first place:

Take quiet, asymmetric, unlikely bets without giving out your position.

For a short seller (someone who bets a stock will fall and tries to profit when its price drops), that can mean everything. Each time his firm would file a disclosure, the media would make their own interpretations, which trickled down to smaller retail traders who like to copy-trade bigger players. That can move a stock irrationally. It made it impossible to operate the way he had all these years. And with that, he closed down the fund and started a new venture: a newsletter!

That’s right. A few weeks into closing down his firm, Burry started his own newsletter, aptly called Cassandra Unchained.

After his newfound freedom from regulatory filings, he started posting on Substack and X with the same warning: the AI boom is looking more and more like a bubble in the making, similar to the dot-com crash of the early 2000s.

His thesis is simple. The companies paving the path to AI — chipmakers, cloud giants, AI-computing companies and of course the infrastructure builders — are investing heavily to build the future. And because the hardware and software are so unique, they are also extremely expensive. Think billion-dollar data centres, massive computing clusters and GPUs whose long-term return on investment is still uncertain at this scale.

He’s even given this a name: ‘supply-side gluttony’. Enormous spending chasing unproven economics.

He argues that many AI-heavy firms are assuming their expensive GPUs and servers will last 5–6 years and are depreciating this hardware over that time frame, when in reality it may become outdated much sooner. And if he’s right, these firms are overstating their profits by as much as $176 billion between 2026 and 2028.

But why does this matter?

To answer that, let’s understand some basic depreciation. Say you run a cab company, and your fleet consists of cars that cost ₹10 lakh a unit. With the heavy use in the city, the car might be good for 5 years of use before its maintenance costs more than it brings in, and also before customers start complaining about it. By that logic, your depreciation costs should be about ₹2 lakh a year.

But on paper, you suddenly claim the cars will last 10 years. Now you only show ₹1 lakh as the annual depreciation cost. The annual cost drops to ₹1 lakh and profits look much healthier, even though nothing has changed in cash terms. It all works fine until year five arrives and the car is actually at the end of its life, but your books pretend it has another five years left. But when the cards come crumbling down, the company has to take a huge hit to its balance sheet just to clean up this accounting.

This is what Michael Burry was getting at when some tech companies suddenly doubled the useful life of their assets for depreciation. Hence the $176 billion figure.

Sure, you could argue that today’s GPUs are built better. But the pace of AI progress is even faster, which means the hardware powering it can become obsolete much sooner than companies assume.

From the outside, it might look like Michael Burry is shorting AI itself because the companies are so large. But he’s really betting against the pricing of the technology behind it. And history shows that big tech shifts don’t always reward investors equally. The dot-com era still has survivors, but most never delivered the returns investors had hoped for.

Similarly, the rate at which current AI companies are building aggressively makes it look unlikely that investors will see returns or earnings to match. And even if they are to match it, earnings will have to grow at exponential rates in the coming years, while their customers have yet to show any sign of durable profits that will sustain such growth. This faultline is what he’s targeting.

Now keeping that in mind, we also have the depreciation accounting factor. Meaning, many AI companies are stretching the useful life of their GPUs and servers, say from 3 years to 6 years. So their annual depreciation expense becomes smaller today. But if revenue drops in the future and investors start questioning these assumptions, companies may be forced to reset the useful life to something more realistic. When that happens, they must catch up on all the depreciation they delayed. That’s the “much higher depreciation costs all at once or a “double hit” that could shatter today’s belief in endless exponential profits.

This is why Burry’s positioning makes sense to him. Shorting these stocks, in his view, puts him in the right place when stretched valuations finally collide with more realistic cash flows.

But does this mean Burry is unquestionably right, you ask?

Well, not everyone agrees. Nvidia, for one, pushed back strongly after Burry’s comments gained traction. The company argued that GPUs and AI servers remain useful for years and simply get shifted to less demanding tasks as they age. Analysts like Bernstein also say the debate isn’t settled yet. And that the real useful life could be longer, shorter, or simply too early to model with confidence.

Yet Burry isn’t waiting for consensus. To him, the best asymmetric opportunities appear before the numbers settle. So realistically, one of two things might happen:

1. AI monetisation and capital expenditure ramp up fast enough to justify today’s prices, giving bullish investors the win, or
2. The pricing is too far ahead of fundamentals, and Burry’s contrarian bet pays off.

Until then, the market will keep leaning one way, and Burry will keep leaning the other, making it a long short that only time can settle.

Fun Fact: In Greek mythology, Cassandra was a priestess who could foresee disasters but was cursed so that no one believed her. After shutting down his asset management company and freeing himself from disclosure rules, Michael Burry named his newsletter Cassandra Unchained — a nod to both the myth and Warren Buffett’s idea of the “Cassandra Syndrome”, where market warnings are dismissed until it’s too late.

Liked this story? Share it with a friend, family member or even strangers on WhatsApp, LinkedIn and X.