Anthropic and Nscale in talks over a $45 billion cloud deal: compute capacity is the new bottleneck

By Anthropic News (via Central Oregon Daily and Bloomberg) · October 19, 2026.

Anthropic and Nscale are in talks over a new cloud computing deal worth around $45 billion, according to Bloomberg News, as picked up by other outlets citing a source. Nscale is an AI infrastructure company backed by Nvidia. These are negotiations: the material we have seen includes no signed contract and no public confirmation from either party.

The available material is thin. It does not include the deal's timeline, the location of the data centers, the type or number of chips, or the planned power capacity. Those details should not be taken as given. The only solid facts are the reported figure and the identity of the parties, both attributed to Bloomberg's source.

To put the number in perspective: $45 billion is a figure on the scale of national infrastructure, not a software contract. More broadly, the major AI labs have spent months signing compute commitments worth tens of billions of dollars with cloud providers and GPU specialists (graphics processors, the chips that train and run the models). This move fits that pattern.

Our take: the thesis is that the limiting factor in AI is no longer just talent or algorithms, but guaranteed access to electricity, chips and data center space. Whoever secures capacity years ahead sets the pace of model releases. Anthropic needs that capacity to compete at the frontier, and diversifying providers reduces its dependence on any single partner.

Who wins: specialists like Nscale, which go from promising newcomer to significant player, and Nvidia, whose customer ecosystem grows wider. Who is exposed: labs with less access to capital, for whom competing at the closed frontier will become ever more expensive. This is an argument in favor of the open frontier, which lowers the cost of access and decentralizes it.

There are short-term risks that should not be sugarcoated: a spending race of this size concentrates power in a few companies, puts pressure on power grids and fuels doubts about whether demand will justify so much investment. The headline about the stock-market rebound of the "AI trade" that accompanies the story on Bloomberg is a reminder that markets swing with every figure.

In the long run, however, this investment is what underpins the deeper promise: models capable of accelerating biomedical research and generating the abundance that frees people. The condition is that compute ultimately translates into useful, accessible capability, not just more data centers. We will follow the closing of the deal to see whether the figure is confirmed and which technical details are made public.