Anthropic, the San Francisco AI lab behind Claude, just locked in a huge deal—$35 billion over six years—with Lambda, a cloud provider backed by Nvidia. The plan is simple: crank up cloud capacity fast enough to keep Claude growing, and do it with some serious Nvidia muscle.
Here’s how the deal is set up: Lambda is getting its hands on a data center in Nueces County, Texas, which Hut 8 is building. If you haven’t heard of Hut 8, it used to mine bitcoin but switched gears into AI data centers. Nvidia isn’t sitting on the sidelines either. They’re leasing the whole Texas facility, supplying the GPUs, and letting Lambda take charge of deploying the hardware. Lambda then sells all that compute power back to Anthropic. In total, the site should deliver around 350 megawatts of capacity.
So why is this important? Basically, everyone in AI is scrambling to grab as much computing power as possible, and deals like this show just how fierce—and expensive—the race has gotten. Anthropic needs to make sure it never hits a bottleneck while training new models or rolling out Claude updates. For Nvidia, it’s a chance to do more than just sell chips—they’re becoming a backbone for the AI world by financing, building, and running these giant projects.
This isn’t even Anthropic’s only major deal this week. Just five days ago, they reportedly signed a $45 billion agreement with Nscale, another big player in AI infrastructure. Anthropic is clearly gearing up for an IPO, and these deals are about securing a tech pipeline that won’t dry up anytime soon.
Bigger picture? The industry is shifting: AI labs are signing massive, long-term contracts with cloud providers so they’re not left scrambling for compute down the road. Data center companies like Hut 8—once in the crypto game—are now flipping their old sites to power AI, because that’s where the demand (and money) is.
For businesses using Claude, this could mean fewer outages and faster product updates, as long as the Texas site goes live on schedule. Still, there’s a catch: stacking so much capacity in just a few big projects might make the AI world a little less diverse and flexible.
Some details haven’t come out yet—like exactly how much Lambda’s paying Nvidia for all this, or payment terms and performance goals. Everyone’s watching to see how fast the Texas center comes online, and whether other AI startups start cutting similar deals to keep up.
