AI Stocks Rally: Real Boom or a Market Crash Waiting to Happen?

If you had invested Rs 10 lakh in Nvidia three years ago and done nothing else, you would be sitting on roughly Rs 1.1 crore today. Palantir has surged over 2,400% in the same period. SanDisk has returned over 4,000% in just one year while Micron posted 800% gains in that time, as per Google Finance.

Now, OpenAI, Anthropic, and SpaceX (merged with xAI) are lining up for what could be the largest cluster of IPOs in financial history. So the question is real and urgent: is this a once-in-a-generation opportunity, or the biggest market bubble the world has ever seen? Let’s break it down.

AI Stocks Rally: Why did Nvidia, Sandisk and others skyrocket?

The rally in AI Stocks began with one simple question: who benefits when every company suddenly wants AI? The first answer was Nvidia. Training an AI model is like checking lakhs of board exam papers across India overnight. A normal computer is like one teacher checking papers manually, it would take forever

Nvidia’s GPUs are like a huge evaluation center where thousands of papers are checked simultaneously. They can handle massive workloads at very high speed. So when every tech giant rushed to build AI models at the same time, everyone needed Nvidia’s chips to do the heavy lifting and the company’s revenue exploded.

Then came the memory stocks. If Nvidia's chips are the brain of an AI data center, Micron and SanDisk supply its memory and storage. You cannot run a brain without memory. As hyperscalers like Google, Microsoft, and Amazon began hoarding chips, SanDisk signed multi-year supply contracts worth over $42 billion. The result: the stock surged more than 4,000% in a year!

The IPO Wave that Could Swallow the US Stock Market

Here is where it gets genuinely complicated. Three AI companies are going public in 2026, targeting a combined valuation of nearly $3 trillion.

Sources: The Wall Street Journal, The Information, Financial Times, VentureBeat, Reuters

The number every Indian investor in US stocks needs to absorb: these three IPOs are expected to collectively demand over $240 billion from public markets. The entire US IPO market raised only $45 billion in all of 2025. SpaceX alone could raise more than every US IPO over the last two years combined. Why is it concerning?

Imagine the stock market as a pond with a steady water level. Now drop three huge rocks in the pond simultaneously. The water gets displaced. Similarly, when $240 billion flows into new listings, institutional money rotates out of existing holdings like Apple, Microsoft, and Nvidia to fund it. Even if you never touch an IPO stock, that rotation could temporarily depress what you already own.

VC analyst Tom Tunguz did the math: at a standard 15% float, these three companies would need to absorb $432 to $576 billion from markets in a single quarter. That exceeds the entire decade of US IPO fundraising.

AI Numbers that do not add up

Traditional software companies like Oracle, Salesforce usually build the product once and then sell the same software to millions of users. After the core product is built, the cost of serving each additional user is relatively low. That is why these companies can command high revenue multiples of 30-40x. AI companies are different. Every new user, prompt, and output requires fresh computing power, which means real cost keeps getting added as usage grows.

Think of traditional software like a recorded cooking video. You shoot it once and sell it to a million people. The cost does not rise much with every new viewer. AI is more like a food truck. Every new customer needs a fresh meal, and that meal needs expensive ingredients. So you cannot value a food truck like a video channel, no matter how good the food is.

OpenAI wants a $1 trillion valuation on $25 billion in revenue, a 40x multiple. That would be fair for a zero-marginal-cost software business. But OpenAI is projected to lose $14 billion in 2026 despite that revenue. Their own internal estimates put cumulative cash burn at $600 billion by 2030, revised up from $550 billion just a month earlier. They are spending more than Rs 2 to earn every Rs 1.

David Cahn, a partner at Sequoia Capital, quantified the industry-wide version of this problem. For every dollar spent on a GPU, another dollar goes to energy and cooling. End-user businesses need a 50% gross margin on top of that just to survive. Total the capital expenditures across the industry and AI needs to generate $600 billion in annual revenue just to break even on hardware. Actual AI revenues today? Roughly $100 billion. The $500 billion gap is being filled by VC cash and corporate debt, not profits.

The Big Risk: How AI Companies are Funding Each Other

Imagine Blinkit offering massive discounts during a sale season. Customers place far more orders because groceries worth ₹500 are suddenly available for ₹300. Order volumes surge, Blinkit’s growth numbers look explosive, investors get excited, and the company raises even more money because the business appears to be scaling rapidly. Blinkit can then use that fresh funding to continue offering more discounts, which again pushes orders higher. So the cycle starts feeding itself. But a part of that growth is being artificially boosted by Blinkit’s own money flowing back into the system through discounts.

That is the concern some analysts have with Nvidia and OpenAI. Nvidia invests billions into OpenAI, and OpenAI uses a huge amount of that capital to buy Nvidia’s AI chips. Nvidia then records those chip purchases as revenue, which boosts its earnings and valuation, giving it even more financial power to invest further into OpenAI. On paper, both are thriving. In reality, one is funding the other to buy from itself. 

It is a documented pattern flagged by researchers at Yale School of Management and INSEAD, and the exact graphic Michael Burry posted on X before announcing his short positions. Yale's Jeffrey Sonnenfeld calls this "circular financing" and warns it creates extreme contagion risk: if end-consumer AI demand ever falls short of projections, the entire ecosystem unravels together, not company by company.

The Man Who Called 2008 Market Crash is Shorting AI

Michael Burry predicted the US housing collapse in 2008 and turned it into a billion-dollar trade. He does not raise alarms for attention. He raises them when he is ready to back them with money. 

“Stocks are not up or down because of jobs or consumer sentiment,” Burry wrote. “They are going straight up because they have been going straight up. On a two letter thesis that everyone thinks they understand. ... Feeling like the last months of the 1999-2000 bubble,” Burry wrote in a Substack post recently.

The Big Short fame’s fund Scion Asset Management currently holds $912 million in put options against Palantir and $187 million against Nvidia. Put options profit when stocks fall. These AI shorts now represent roughly 80% of his entire portfolio. That is not a hedge. That is conviction. 

He has additionally expanded bearish bets to include Oracle, the iShares Semiconductor ETF, and the Nasdaq-tracking QQQ, with expiries into 2027. He is not betting one company will stumble. He is betting the whole ecosystem resets!

Read More: Michael Burry’s Bet Against AI Justified? Logic Explained

Burry is not alone. Amazon founder, Jeff Bezos has called the infrastructure build-out an "industrial bubble." Goldman Sachs CEO David Solomon expects "a lot of capital deployed that doesn't deliver returns." Researcher Julien Garran at MacroStrategy Partnership estimates the AI bubble is 17 times larger than the dot-com bubble and four times bigger than the 2008 real-estate collapse.

What Should Indian Investors Do?

The AI trade is not one thing. It is several things layered on top of each other with very different risk profiles. Infrastructure stocks like Micron and SanDisk have real profits, real contracts, and forward valuations that are still reasonable after the rally. The revenue is real; the customers are hyperscalers with genuine demand. AI software stocks like Palantir are trickier. Real product, real revenue, but Palantir trades at over 100 times sales. That bakes in years of flawless execution with no margin for error. It is already down more than 34% from its 52-week high.

The IPO layer is the most speculative tier. Trillion-dollar valuations on companies burning cash by the tens of billions, in an industry still $500 billion short of breaking even. If you are investing in US Stocks from India, the single most important number to watch is not revenue growth or user counts. It is whether AI revenue is finally closing the gap with infrastructure spend. Until that gap narrows, these valuations rest on narrative, not evidence.

Michael Burry sees a cliff. Sam Altman sees a runway. The truth is, nobody knows which one AI becomes yet. So don’t ignore the opportunity. But don’t ignore the risk either. In this market, blind optimism and blind fear can both be expensive.