Episode Details

This week we talk about OpenAI, the Stargate Project, and Meta.

We also discuss o1, AGI, and efficiency.

Recommended Book: The Shortest History of Economics by Andrew Leigh

Transcript

One of the bigger news items these past few weeks, in terms of the numbers involved, at least, was an announcement by US tech company OpenAI that it will be starting a new company called the Stargate Project, which will boast a total $500 billion-worth of investment, the first $100 billion of which will be deployed immediately.

All that money will be plowed into artificial intelligence infrastructure, especially large-scale computing clusters of the kind required to operate AI systems like ChatGPT, and the funds are coming from OpenAI itself, alongside SoftBank, Oracle, and MGX, with Arm, Microsoft, and NVIDIA also involved as technology partners.

It’s a big, beefy enterprise, in other words, and the fact that this has been in the works since 2022, it’s official announcement seemingly held back so that newly returned US President Trump could announce it as part of his administration’s focus on American infrastructure and AI dominance, didn’t dim the glow of the now-formal announcement of what looks to be a truly audacious bet on this collection of technologies, doubts about the players involved having the money they’ve promised ready, notwithstanding.

That said, this is far from the only big, billions and tens of billions-scale wager in this space right now.

Last year, Microsoft announced a $30 billion infrastructure fund, in collaboration with BlackRock, and earlier in January of 2025, Google’s CEO said that his company would spend about $80 billion on the same, separate from their commitment to Stargate.

Meta’s CEO Mark Zuckerberg recently divulged that the company would spend somewhere between $60-65 billion on capital expenditures, mostly on AI, in 2025—that’s up about 70% from 2024 spending.

And last December, xAI CEO Elon Musk announced that his company had just raised a fresh $6 billion to build-out more compute infrastructure; and his role at the head of that company is assumed to be part of why he trash-talked the aforementioned Stargate effort, though there’s also a long-simmering animus between him and OpenAI CEO Sam Altman, and the fact that everyone seems to be trying to get in good with Trump—which is probably part of why many of these announcements are happening right now: Trump is in the position to king-make or cripple their respective efforts, so whomever can get in good with him, or best with him, might have an advantage in what’s become a very expensive knife-fight in this most rapidly burgeoning of tech investment loci.

There’s a reason there’s so much money flowing to this space, announcements aside, right now, too: the chatbots that’ve emerged from the GPT, LLM era of AI systems are impressive and useful for many things, and AI powered bots could even replace other sorts of user interfaces, like search engines and apps, with time.

But there are also some more out-there efforts that are beginning to bear fruit.

AI is helping Google’s DeepMind team discover new materials at an astonishing rate—including both the discovery and the testing of their properties, stage.

AI systems are also being used to accelerate drug discovery and trial design, and a company (backed by OpenAI’s Altman) is trying to extend human life by a decade using exactly this process.

Meta has a new tool that enables real-time speech and text translation between up to (depending on the type of translation being done) 101 different languages, and we’re even seeing AI systems meant to detect and track small, otherwise overlooked infrastructure issues, like potholes, at a local level.

And to be clear, this is far from a US government and US-based tech company effort: g