Investigating the Overlooked
Before there was a company called Oracle, there was a CIA project called Oracle. Larry Ellison, Bob Miner, and Ed Oates worked on it — a data-management effort with roots in a contract at Ampex — and when they left in 1977 to found what became Oracle, they took the project's codename with them, with permission, and made the intelligence community their first customer.[1] Half a century later, Oracle runs federal and defense cloud infrastructure and this month declared force majeure on a Stargate AI data center it hasn't finished.[2] Across that entire arc, the US government has owned exactly none of it. That is not an oversight. It is how America is built.
Start with the part that is easy, because naming the easy case is what makes the hard one visible. In a state-capitalist system, the question of where government-created technology should live has a trivial answer: it goes into the company the state owns. China seeds a technology and it flows into a national champion — a firm the government controls, directs, and profits from. The IP has a natural home, because the state is already an owner of industry. The vessel exists before the invention does.
The United States, in general, does not have state-controlled industry — and has spent most of its history treating the idea as faintly un-American. That single fact creates a structural problem no autocracy has to solve: the American government is one of the most prolific creators and funders of frontier technology on earth — the CIA, DARPA, the national labs, NIH, the university system — and it has no natural place to put what it makes. It invents, and then it must hand the invention to someone else, because it owns no company to hold it in.
So it gives it away. That is not a slur; it is the codified law of the land. The Bayh-Dole Act of 1980 settled the question for federally funded research: the government pays for the invention, and the university or contractor keeps the patent and commercializes it.[3] The taxpayer funds the R&D and keeps the receipt; the private entity keeps the asset. DARPA seeded the internet, GPS, and the guts of the modern smartphone, and the value accrued to private companies, exactly as designed. Oracle is just the earliest and cleanest emblem: government origin, government first customer, private ownership forever.
There is one place America built something closer to a vessel, and it is the part I know from the inside. In 1999 the CIA chartered In-Q-Tel as a nonprofit precisely so it could take equity in the companies whose technology it needed — a workaround invented because the government had no lawful way to own a stake and no company to hold one in. It is the exception that proves the rule: the US had to invent a special vehicle to do the thing a state-capitalist system does by default. Even then, In-Q-Tel holds a stake in a private company. It is a foothold, not a home.
Which is why the last two years look so strange against two centuries of precedent. Washington has started taking direct equity — in chipmakers, in critical-minerals firms, in a run of quantum companies — and floating a sovereign wealth fund. Read against the Oracle story, that is not a lurch toward socialism so much as a country improvising, under competitive pressure, the vessel it deliberately never built. The government that gave Oracle away is now trying to figure out where to put the next one.
That is the question sitting under the whole American system, and it does not have a settled answer: when a democracy that refuses to own industry nonetheless creates the most valuable technology in the world, where is it supposed to put it? Give it to the private sector and you keep your freedom and lose the asset. Hold it and you become the thing you defined yourself against. America has always chosen the first door. It is only now, watching a rival that chose the second, starting to wonder about the cost.
Part of the same thread: "Oracle Invoked Force Majeure… Google's Fix Is to Put the Data Center in Orbit."