Observing a 25-year-old become a billionaire in a matter of weeks has a certain surreal quality. Michael Truell, the founder of Anysphere, a business that most people outside of a small tech corridor had never heard of, saw his startup valued at $9.9 billion in June. Shortly after, he reportedly received offers that increased that figure to between $18 and $20 billion.
He may now be worth more than the yearly public education budgets of the majority of nations. Additionally, it’s still unclear if the product is worthwhile. However, investors appear to think so, and the market is currently their belief.
| Category | Detail |
|---|---|
| Total AI unicorns (private, $1B+ valuation) | 498 companies, combined value $2.7 trillion (CB Insights, 2025) |
| New unicorns founded since 2023 | 100+ companies reaching billion-dollar valuations in under two years |
| Anthropic current valuation (in talks) | $170 billion — nearly 3× its March 2025 valuation |
| OpenAI proposed secondary valuation | $500 billion, following a $300 billion fundraise round in early 2025 |
| Thinking Machines Lab (Mira Murati) | $2 billion seed round — largest in history — giving it a $12 billion valuation |
| Nvidia market cap milestone | First company globally to reach $5 trillion in market value (October 2025) |
| Elon Musk net worth (year-on-year change) | Rose ~50% to $645 billion; first individual to exceed $500B net worth |
| San Francisco billionaire count vs. New York | |
| 73 events including M&A, IPOs, and corporate stakes (CB Insights) | |
| Wealth creation assessment (MIT) | “We have never seen wealth created at this size and speed” — Andrew McAfee, MIT principal researcher |
The peculiar rhythm of the AI boom is that valuations are coming in after a pitch deck and a demo rather than years of revenue and profit. This is not wholly novel. This reasoning was applied differently during the dot-com era. However, this time, the scale and speed feel really different.
In terms of wealth creation velocity, nothing can match what AI is producing, according to MIT researcher Andrew McAfee, who examined more than a century of economic data. Given that it encompasses the Gilded Age, the railroad boom, and the internet, that assertion is worth pondering for a while. Everything.
It’s difficult to miss the shift if you stroll through San Francisco today. A few years ago, this city was being written off—empty office floors, shuttered stores, headlines about a “doom loop” that seemed frighteningly real. More houses sold for more than $20 million last year than in any previous year on record, according to Sotheby’s International Realty.
Over the past ten years, the number of millionaires in the Bay Area has doubled, surpassing the 45% growth in New York. There are currently 82 billionaires in San Francisco compared to 66 in New York City. Once again, the cafés are packed. The rent is exorbitant.
With a combined estimated value of $2.7 trillion, there are currently 498 private AI companies worth $1 billion or more. One hundred of those businesses were established after 2023. Consider this: some of these companies are already worth more than mid-sized economies, and they are hardly old enough to have completed a full hiring cycle. Over 1,300 AI startups are valued at more than $100 million.
This is not a slow change in an industry. It appears more like a tectonic, abrupt geological event that changes the terrain before anyone has had time to adapt.
In September 2024, Mira Murati departed OpenAI. She started Thinking Machines Lab in February 2025. At a $12 billion valuation, she raised $2 billion by July, the biggest seed round in history. It’s difficult to avoid feeling that the standard guidelines for company formation have been suspended as you watch this develop.
The goal of a seed round is to validate a concept. $2 billion isn’t proof-of-concept funding. It’s money from the whole industry. It implies that investors are placing bets on who built these businesses rather than on what they have created.
According to reports, Anthropic is in negotiations to raise $5 billion at a valuation of $170 billion, which is almost three times its value just a few months ago. Even on paper, CEO Dario Amodei and the other company founders are probably multibillionaires. In order to give cash to employees who own equity they haven’t been able to access, OpenAI is investigating a secondary share sale at a proposed $500 billion valuation.
This boom differs from previous ones in the structure of wealth creation. Instead of publicly traded shares, the majority of it is locked in private equity. In theory, people are extremely valuable, but they are practically unable to spend much of it.
Secondary markets are starting to alter that. Tender offers, structured share sales, and the capacity to borrow against private equity stakes are all subtly creating channels of liquidity that were nonexistent ten years ago. Lucy Guo, a co-founder of Scale AI who left the company in 2018, reportedly paid about $30 million for a mansion in the Hollywood Hills after Meta invested $14.3 billion in the company.
That’s what it looks like in 2025 to successfully leave a company you left six years ago. Even those who were not present in the room can be affected by a single transaction.
A version of this tale is told through public markets in Jensen Huang’s year. Nvidia became the first company in history to reach a $5 trillion market capitalization, which, depending on the metric, is greater than the GDP of either India or Japan. In just one year, Huang’s personal wealth increased by almost $42 billion. Along the way, he also sold off shares worth almost $1 billion.
These days, Nvidia’s chips are something like the oil of the AI economy; without them, data centers cannot function, models cannot be trained, and valuations cannot fluctuate. This unsettling concentration of vital infrastructure in one organization raises issues that regulators haven’t yet addressed.
Last October, the Bank of England quietly warned that equity market valuations for AI-focused businesses seem to be stretched by a number of factors. The bank warned that a sudden correction could have an impact on international markets if investor confidence turns out to be misplaced. This type of language falls under the category of “things everyone already suspects but nobody wants to say too loudly.”
The dot-com analogy keeps coming up because the emotional architecture seems familiar, not because the current boom is the same. There is a belief that this technology transforms everything, and it is currently bearing a heavy financial burden.
However, there is a structural difference between AI and the internet boom of the late 1990s. The goods are authentic. Some of these businesses have actual revenue. Data centers, power grids, and custom chips are all part of the infrastructure buildout, which requires billions of dollars in capital expenditures and generates employment, contracts, and tangible assets.
Unlike pets.com, it’s not entirely speculative. The question is not whether AI is beneficial. It concerns whether these specific businesses, at these specific valuations, will be profitable in the upcoming ten years. There is currently no definitive answer to that more difficult question.
Silicon Valley is still Silicon Valley, according to McAfee. People have been forecasting its replacement for the past 25 years: Shanghai, Austin, Miami, and London. It didn’t stick. The AI wave has only strengthened the concentration of people who know how to start, finance, and grow technology companies in a single Northern California metropolitan area. He remarked, “It’s amazing how geographically limited all of this still is.” The wealth is enormous and worldwide in scope. There are surprisingly few zip codes.
