The biggest news today is Stargate. President Trump, flanked by Softbank founder (and legendary Japanese tech billionaire/investor) Masayoshi "Masa" Son, Oracle chairman Larry Ellison, and OpenAI CEO Sam Altman, announced something called "Project Stargate," a $500 billion (that's 5, 0, 0, billion dollars) project to build the biggest AI-training facility—really the biggest computer in history—in Texas.

This was hailed as a Project Manhattan-type investment by the US government to build an artificial general intelligence, and even an artificial superintelligence. If you're not familiar with the lingo, artificial general intelligence or AGI means AI that's as smart as the smartest humans—which we are tantalizingly close to. Artificial superintelligence, or ASI, represents AI that is vastly more intelligent than even the smartest humans, and in some scenarios, an ASI would be capable of improving its own code, so that it would keep getting more and more intelligent potentially with no upper limit. These are difficult concepts to grapple with, as we are strangely close to science-fiction territory. How should we think about an artificial superintelligence? In some sense, the question is self-refuting, since, whatever an ASI ends up being (if it does end up being), that would be, by definition, beyond our understanding. We would really only have two options: accept its hopefully-benevolent rule, or try to destroy it.

Until then… Until then, AI clearly has very important uses in the fields of technology, healthcare (as Ellison emphasized during his turn at the lectern) and, especially, national security. And the best Chinese AI, DeepSeek, according to experts on X and experts your correspondent spoke to, is close to catching up to the best American models, and progressing fast.

And this must be clarifying. Whatever the risks of an American AGI or ASI, surely they are less than the risks of a Chinese Communist AGI or ASI.

So: Manhattan Project! Apollo! Gotta beat the commies!

Except that Stargate is not at all a Manhattan Project. For starters, as presented, 0 dollars or direction are coming from the Federal government. This is good news from a taxpayer or pro-free market perspective, to be sure, but that's not what people usually mean when they say "Manhattan Project."

Secondly, where are the $500 billion going to come from? That is a very large number, even for the players involved. A hundred billion here, a hundred billion there, and pretty soon you're talking about a lot of money… Softbank, according to a quick search, has $347.7 billion under management to invest in all tech projects it plans to invest in. Oracle, meanwhile, is not an investment firm, and has roughly $11 billion cash on hand. Elon Musk, whose personal hatred for Sam Altman is notorious, tweeted in response to the announcement that he has it "on good authority" that the Stargate joint venture only has "well under" $10 billion "secured" out of the announced $500 billion. Regardless of the reliability of Musk's information, which is not always perfect, he is surely roughly right: Stargate doubtless has, as of this writing, several billion dollars on hand to start building a facility, and plans to raise the rest of the $500 billion in the future. We can be generous and say that it has a good chance of being able to raise such a large amount. But there is no guarantee. The $500 billion figure, as of now, is just a press release.

Thirdly, what about the timing? During the announcement, Masa enthused publicly to Donald Trump that they had signed the papers "yesterday" because such a thing could only have been possible during a Trump Administration. Then during an interview with Fox News, to emphasize the reality of the project, Larry Ellison alleged that they had already began construction of the facility in Texas, which couldn't be true if they'd formed the joint venture just the day before. We are not accusing anybody of lying. They may both be technically correct: the players may have signed some additional or final legal document the day before. But the Stargate project, or something like it which may or may not have been rebranded or expanded into Stargate, clearly predates it.

So, to be clear, Stargate is not some great new Trumpian and Federal government Manhattan Project for AI breakthroughs. It is a private investment project, which has doubtlessly been in the works for months, and which got a rebrand and a facelift to create a photo op and a press release with President Trump. Nothing wrong with that. But readers of PolicySphere should just be clear about what's going on.

Now that this has been said, putting aside the hype, what remains is that a coalition of tech titans, including OpenAI, whose models are the best or second-best on the planet, are going to build the biggest supercomputer in history, in order to try to create AGI or something close to it. This is still a very significant development.

From a policy perspective, the only takeaway is this (and this should be fodder for another article): they're going to need a lot of energy, and the best way to do that is to legalize building lots of small modular nucelar reactors!