770 People Started a Wealth Transfer. Here’s Where It Went.

4,500 People vs. 191,000 People

In late 2022, a 770-person company released ChatGPT. That single product ignited the AI boom and sent tech market caps soaring by trillions. Three years later, that company — OpenAI — has about 4,500 employees and a valuation of $840B.

At the same moment, Alphabet has approximately 191,000 employees and a market cap of roughly $3.7T.

Divide enterprise value by headcount:

Enterprise value: public companies at March 2026 market cap, private companies at most recent funding round. Headcount: latest public filings.

OpenAI's value per employee is $187M. About 1.6x NVIDIA, 10x Alphabet, 125x Amazon. Anthropic sits at $127M, outpacing most of Big Tech. By the density of value generated per person, AI startups are in a class of their own.

But look at the rightmost column. The story inverts.

The Reversal

Sam Altman's net worth is approximately $2B. He runs an $840B company and is worth $2B because the answer is simple: he has no equity in OpenAI. His wealth came from his YC-era portfolio — Reddit, Stripe, and similar investments. His CEO salary is $76,000. The man who triggered the AI revolution earns less than a junior engineer at a Silicon Valley startup.

Larry Page, meanwhile, is worth $257B. Alphabet's value per employee is $19M — one-tenth of OpenAI's — but Page's personal wealth is 128x Altman's. Jensen Huang is at $154B. NVIDIA's value per employee is roughly two-thirds of OpenAI's, but Huang's wealth is 77x Altman's.

The pattern: the organization generating the most value per person has the founder with the least personal wealth. The lower the value density, the larger the founder's fortune. Plotted on a graph, value per employee and founder net worth are inversely correlated.

The Structure That Created the Reversal

Ownership structure is the obvious explanation. Page, Zuckerberg, and Huang have held equity for decades. Altman started as a nonprofit and has none. But "he has no equity so he's not rich" just restates the table. The more important question is which direction value flows.

OpenAI built ChatGPT. Who got richest? Jensen Huang, whose company supplied the GPUs ($20B → $154B, +670%). Larry Page, whose company jumped into the competition ($111B → $257B, +132%). Mark Zuckerberg, who responded with Llama ($67B → $222B, +230%). Value flowed not to whoever created the disruption, but to whoever owned the infrastructure the disruption runs on.

And these infrastructure owners enrich each other in a closed loop. NVIDIA makes GPUs. Google, Meta, Amazon, and Microsoft buy them. As those companies scale their AI services, GPU demand rises and NVIDIA stock goes up. One side getting richer makes the other side richer.

OpenAI's most recent $840B funding round makes this structure explicit. The investors: Amazon ($50B), NVIDIA ($30B), SoftBank ($30B). The companies funding the AI creator are the very infrastructure owners who got rich from AI. Value generated by innovation accrues to the infrastructure owners, and that capital turns around to buy the next wave of innovation.

Is the Reward Really Not Going to Innovators?

A pause is warranted. Is this reversal permanent, or a timing issue?

Reports indicate Altman could receive roughly 7% equity in OpenAI when the for-profit conversion completes — though he has publicly denied specific plans. At current valuations, that would be approximately $59B, which would put him in the top 20 wealthiest people in the world overnight. If Anthropic goes public from its $380B valuation, Amodei's wealth could jump substantially. What looks like a reversal now might simply be the reward that hasn't arrived yet.

Worth noting too: private company valuations aren't directly comparable to public market caps. OpenAI's $840B was set in a funding round, not traded daily. A significant portion of that round was structured as compute credits and conditional tranches rather than cash.

Still, one thing holds. Even if the reward "eventually" arrives, it arrives after infrastructure owners' wealth has already compounded for years. The sequence matters. Innovators get paid last. Infrastructure owners get paid first. That ordering itself has structural effects.

The Question This Raises

The Gold Rush analogy gets used often: the people selling pickaxes made more money than the miners. But pickaxes in that era could be made by anyone. Today's pickaxes — GPU design capability, global cloud infrastructure, decades of accumulated data centers — can't. For the analogy to hold in 2026, you have to imagine a world where only a handful of pickaxe factories exist, their owners are each other's best customers, and they're also investing in the miners.

One more thing catches. When AI startups eventually exit through IPO or acquisition, who captures the proceeds? Early investors and some employees — but the largest checks go to Amazon ($50B in) and NVIDIA ($30B in). Even when the innovation's fruits "eventually" come around, the biggest payouts still go to the infrastructure owners. The sequence doesn't change. It repeats.

So what's the real position of people building innovation in this era? Independent companies, but dependent on infrastructure owners for capital, compute, and exit. Less an independent innovator and more a portfolio company of the infrastructure ecosystem.

$76,000 per year is leading an $840B revolution. The moment that sentence reads not as a testament to idealism but as evidence of a structure — that's when the questions start.