Meta’s AI appetite keeps the neoclouds busy

Meta’s not done shopping for compute

Meta Platforms is leaning on neocloud providers CoreWeave and Nebius Group for AI computing capacity, which is corporate-speak for: the company wants a lot more GPU muscle, and it wants it now.

That matters because AI infrastructure is turning into the new arms race. The bigger the model ambitions, the bigger the tab for chips, data centers, power, and all the plumbing that keeps the whole thing from melting into a very expensive science project.

What this means for investors

If you own Meta, the takeaway is both simple and slightly annoying:

• Meta still has to spend big to keep up in AI.
• Outside providers like CoreWeave and Nebius can win business when hyperscalers need capacity fast.
• The flip side is that Meta is also building its own AI stack, so these deals may be more about bridging a supply gap than outsourcing the future.

That makes this a double-edged update. Meta’s AI push looks serious, but serious also means costly. And for neocloud names, it’s another validation that the market for spare AI compute is still hot enough to fry an egg on it.

Big picture

This isn’t a one-company story in the long run — it’s a snapshot of the AI infrastructure pileup. Everyone wants more compute, everyone wants it yesterday, and the companies that can deliver it are suddenly very popular dinner guests.