Nscale, one of the fast-rising AI cloud providers, says it now holds more than $100 billion in contracted revenue after landing Anthropic as a customer. That figure comes from an exclusive report by The Information, which detailed how the infrastructure startup is using the deal to signal it belongs in the same conversation as the biggest names in AI compute. For a company that most people outside the data center world have never heard of, that’s a staggering number to put on the table.
Here’s why it lands the way it does.
What actually happened
Nscale rents out the raw compute that AI labs need to train and run their models: racks of GPUs, power, cooling, and the networking to tie it all together. According to The Information, the Anthropic win is the anchor behind Nscale’s claim of $100 billion-plus in contracted revenue. Contracted revenue isn’t cash in the bank. It’s the total value of deals signed over their full lifetime, often stretching many years. Still, a backlog that size tells you customers are locking in capacity far into the future.
Anthropic matters here for a specific reason. It’s one of the few labs with the model demand and the funding to sign compute commitments at this scale. When a lab like that picks a provider, it’s a vote of confidence that pulls in more customers and more investor money.
Why this matters for the industry
A new class of companies has emerged over the past two years, often called neoclouds. These are specialized providers built almost entirely around AI workloads, unlike the general-purpose giants like AWS, Azure, and Google Cloud. CoreWeave is the name most people know. Nscale is now positioning itself as a serious European-rooted contender in that same category.
What stands out is the speed. The status quo not long ago was simple: if you wanted serious GPU capacity, you went to a hyperscaler or you waited. Now labs are spreading their bets across newer providers to secure supply, avoid single-vendor lock-in, and get better pricing. That shift is exactly what lets a company like Nscale rack up a nine-figure backlog in a short window.
A few things to keep in mind about numbers like this:
- Contracted is not collected. The $100 billion reflects future commitments, not current earnings. Delivery depends on Nscale actually building and powering the capacity it promised.
- Concentration is a risk. Leaning heavily on one marquee customer cuts both ways. It’s great validation, but it also means a lot rides on that relationship holding.
- Power and hardware are the real bottleneck. Signing deals is one thing. Securing enough electricity, land, and GPUs to fulfill them is the harder problem across the whole sector right now.
The bigger picture
This is another data point in the compute land grab that’s defining the AI economy. Labs need more capacity than the market can currently supply, so they’re signing long-term deals with anyone who can credibly deliver. Providers, in turn, use those contracts to raise capital and build even more. It’s a flywheel, and it’s spinning fast.
The risk baked into all of it is straightforward. These commitments assume AI demand keeps climbing at its current pace. If model economics shift or demand cools, some of that contracted revenue could look a lot softer than it does today.
What to watch next
Expect more of these announcements as neoclouds compete to prove they’re bankable at scale. The signals worth tracking: whether Nscale can actually deliver the capacity behind its backlog, whether it adds more anchor customers beyond Anthropic, and how it funds the buildout. For the full breakdown, the original reporting is at The Information.