What's happening at OpenAI meanwhile?
Continuing from the Anthropic post — a contrasting picture from their main competitor. Last week CNBC released an inside scoop that OpenAI revised infrastructure spending plans through 2030 — from $1.4 trillion down to $600 billion.
Minus 57% from original plans.
In parallel, Stargate didn't take off — that same $500B megaproject everyone was talking about at the start of last year. A year later, the three partners (OpenAI, Oracle, SoftBank) still haven't hired people or built a single data center. Essentially, they just can't agree on terms.
Many took the news as a sign that "this ugly AI bubble will finally burst." But it seems to me there's still plenty of money in the AI industry. Investors are just tired of presentations about infrastructure megaprojects that only burn billions and deliver nothing, even on paper.
Now everyone wants to see products with a clear use case that generate real revenue today. Anthropic understood this first — and while competitors explain why their Excel plans didn't work out, they're simply building the best developer tool (who knew that was such a profitable niche). And capturing billions in ARR.
OpenAI meanwhile crossed the 910 million WAU mark, projects revenue of $50B in 2026 and $250B by 2030. Consumer subscriptions and enterprise contracts will contribute roughly equally. Personally, OpenAI's strategy raises questions for me. Though their Codex coding agent is actually excellent. We'll see!