Venture capitalists and governments have poured nearly a trillion dollars into AI. That investment accounts for 92% of stock market growth in 2025. Nvidia (NASDAQ: NVDA) alone , the chipmaker powering the boom is now worth $4.5 trillion and represents 45% of NASDAQ growth.
The bet is simple: superintelligence is coming, and whoever builds the infrastructure first wins infinite riches.
There’s just one problem. The math doesn’t work unless a literal miracle happens.
The Holy Grail Problem
Listen to any AI booster out of Silicon Valley, and you’ll hear the same pitch: AI will cure cancer, unlock fusion energy, predict financial markets, and create better art than any human ever could. Then it will improve itself exponentially until it becomes something close to divine.
Sam Altman calls it inevitable. Mark Zuckerberg says he’s prepared to “waste a couple hundred billion dollars” chasing it. The premise is that if you build too slowly and superintelligence arrives in three years instead of five, you’re out of position on “the most important technology in history.”
This is not how economic paradigm shifts have ever worked. The cotton gin, the printing press, and early computers all started with incremental investments that paid off with incremental profits over decades. AI investors are doing the opposite: calling the home run before swinging the bat.
The Circular Money Problem
Nvidia’s rise looks a lot more precarious when you examine where the money is actually coming from.
A Bloomberg chart shows the flow of capital in and out of Nvidia, and it’s moving in a circle. Nvidia invests billions in data center companies. Those companies spend that investment buying Nvidia chips. The same chunk of money gets passed back and forth, claimed as investment, asset, and revenue, sometimes all three.
This only makes sense if AI’s eventual profitability is so vast that these investments are just short-term bridges to capture future gains. But those gains need to materialize fast. AI chips have a lifespan of roughly three years. The trillion-dollar investment needs to earn out before those chips become worthless.
The $800 Billion Question
OpenAI posted $13 billion in annualized revenue this year ,the largest in the AI services space. That’s real money. But Sequoia’s David Kahn estimates AI companies need to sell $800 billion worth of services over the life of today’s data centers and GPUs just to break even. Bain & Company puts the 2030 target at $2 trillion in revenue.
The only way to hit those numbers is an exponential growth curve starting now. Ask yourself: is your own use of ChatGPT likely to boost your productivity 15-30x over the next three to five years? That’s what the investment thesis requires.
The Horny Chatbot Tell
In October, Sam Altman announced OpenAI would “treat adults like adults“, opening the door to customized erotica.
As one internet commenter put it: “If I genuinely believed I was 18 months away from superintelligence that could solve cancer, I probably wouldn’t be pivoting to horny chatbots.”
The dream is cracking.
Four Scenarios, One Likely Outcome
Here’s how the AI boom ends:
Utopian: AI achieves superintelligence, transforms the economy, and makes humanity exponentially more productive. Requires a miracle within three years.
Dystopian: AI goes rogue, makes humanity obsolete. Can’t happen without superintelligence first — which isn’t happening on this timeline.
Integrative: AI fails to achieve superintelligence but solves some productivity problems like any other technology. Modest gains, no exponential returns.
Failure: Investment collapses. Data centers become worthless. Early investors already cashed out.
The early investors don’t need superintelligence to win. They already have. Nvidia traded at $5 in January 2018. Today it’s $180 , a 36x return. Anyone selling the dream along the way profited off a future that never needed to exist.
The Date
If superintelligence is coming, we’d need to see exponential revenue growth within the next 18 months, half the runway before current chips lose their value.
April 1, 2027.
If it’s not obvious by then that AI is really the future, there’s a good chance it was an April Fool’s joke all along.
The rails are built. The question is whether anyone’s actually going to ride them or whether we’re all just watching the same money move in circles until the music stops.
This article is for informational purposes only and does not constitute investment advice.
