Last week, Nvidia announced a profit of $19.3 billion, but only $14.5 billion showed up as cash. A $4.8 billion vanishing act. In chip manufacturing—a business that measures success in atoms—TSMC and AMD convert more than 95% of their profit into cash. Nvidia converts around 75%.
This is not a scandal. This is not a judgement on Nvidia, which makes astonishing technology. But it is the kind of discrepancy that tells you something about the world around this industry. It reveals the oddity of an industry in which demand is often not created by people who need things but by people who owe each other favours, credits, and existential validation.
Consider the circular river in which the same dollar seems to swim across five companies and return, somehow heavier. Nvidia gave $2 billion to xAI. xAI then borrowed $12.5 billion to buy Nvidia chips. Microsoft invested $13 billion in OpenAI. OpenAI committed $50 billion to Microsoft’s cloud. Microsoft placed a $100 billion Nvidia order for that cloud. Oracle granted OpenAI $300 billion worth of cloud credits, prompting OpenAI to order still more Nvidia chips for Oracle data centres.
The Mirage of Profit
Everyone records revenue. No one seems to pay the full bill. The money behaves like an overworked extra in a mythological TV serial— appearing in many scenes, wearing many costumes, always looking strangely familiar.
To be fair, the AI revolution is real. But revolutions attract opportunists. And this one has attracted ‘corporate visionaries’ who have discovered the joys of committing money they don’t have, to buy things they don’t need, to produce services for customers who don’t yet exist.
Airbnb’s CEO recently described these revenue claims as “vibe revenue,” a phrase that should be carved on the gates of Silicon Valley. OpenAI burns $9.3 billion a year while earning $3.7 billion. Yet it is valued at $157 billion - a number whose implied future profits would make even the most enthusiastic astrologer blush. MIT researchers estimate that 95% of enterprise AI deployments fail to produce sustained value. It turns out that “write me a poem about my boss in the style of Rumi” is not, by itself, a business model.
Markets, too, are beginning to move with the slow, graceful decisiveness of predators. Peter Thiel sold $100 million worth of Nvidia stock on November 9. SoftBank sold $5.8 billion soon after. Michael Burry, the man who stared down Wall Street in 2008 has placed a bet that Nvidia will fall to $140 by March 2026. But these rare anyone’s guess, and someone always profits from bets.
Bitcoin, which has become the speculative shadow of the AI sector, has dropped from $126,000 in October to below $90,000. AI startups, who pledged $26.8 billion in Bitcoin as collateral, now sit in the curious position of being technology pioneers, whose fate depends on a digital asset invented to bypass the financial system they’re now entangled in. If Nvidia falls another 40%, cryptocurrency trackers feel that forced liquidations could push Bitcoin to around $50,000.
Betting on Future Profits with Today’s Losses
Those who are speculating on the future of next quarter in Corporate America are predicting a timeline that reads like a screenplay. This is only what some quarters are predicting and not a commentary about the company.
February 2026: Nvidia will report how many unpaid invoices are now older than 60 days.
March: credit-rating agencies take an interest.
April: the first restatement.
The whole structure of eighteen months of exuberance might well unwind in ninety days. The fair value some analysts whisper about is approx $70 a share, far from current loftier levels.
Anyways, this is not about the company. For none of us can predict the future.
AI's Technology Faces Financial Reality
There are lessons here, uncomfortable but useful. About the entire AI rush, valuations, and our own FOMO about investing in AI.
The first is that cash remains the only adult in a room full of prodigies. Revenue is a claim; profit is an opinion; cash is the settlement of all arguments. When profit and cash diverge, it means the real world is gently refusing to agree with the much distributed PowerPoint.
The second is that AI’s commercial foundation is still forming. It is a magnificent technology, but it is not yet a magnificent business. Much of the “demand” comes from within the ecosystem itself - companies buying from each other because everyone is terrified of being left behind in a race they don’t fully understand or that they are building as the story to sell to the world.
The third lesson is about valuations. When a company losing $5.6 billion a year is valued for trillions in future profit, markets are performing a collective ritual of belief.
The fourth is that circular-capital is not value-creation. When the same dollar produces multiple revenue entries across multiple companies, it is not growth.
And finally, there is the lesson that belongs to our time: in advanced economies, the regulatory architecture continues to jog while the market sprints. When these markets cheer in the name of deregulation, just for global supremacy it wants to gain from AI, and from the distraction it creates for world to dance to its tunes, who will pay the bill when the music stops?
Yet amid all this, it must be said: AI is the most consequential technology of our lifetime.
The danger does not lie in AI itself but in the financial circus about it and around it. Every revolution undergoes this moment when the narrative meets the numbers. And oddly, that is when revolutions truly begin. Ain’t it?
(Srinath Sridharan is Author, Policy Researcher & Corporate Advisor, Twitter: @ssmumbai.)
Views are personal, and do not represent the stand of this publication.
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
