Anthropic reportedly in talks on a $45 billion Nscale expansion, spotlighting Nvidia’s role in AI infrastructure

Anthropic is reportedly in discussions about a massive expansion plan that would raise its spending footprint by about $45 billion, according to a market report published this week. The same report suggests Nvidia could benefit if Anthropic’s next phase translates into additional purchases of AI hardware and related infrastructure.

The proposal is tied to Nscale, described in the report as part of Anthropic’s growth plan. Nscale is presented as an effort to scale capacity for training and serving AI models, an area that can quickly become capital intensive when demand for frontier models accelerates.

The market report also frames the timing as important because Anthropic is widely expected to pursue an initial public offering in the near term. If Anthropic is preparing for an IPO, a large, pre-IPO buildout could affect both its capital needs and the optics investors look for around cost discipline and growth velocity.

For Nvidia, the logic in the report is straightforward: AI model scaling tends to translate into more demand for accelerated compute. Nvidia’s GPUs and its broader data-center platform are commonly used by companies building large AI systems, so a major capacity expansion by a major AI developer can pull forward infrastructure orders.

What is not clear from the published report is the specific structure of any potential Nscale expansion, including whether it would be funded through new equity, debt, or a mix of sources. The report also does not lay out who would supply the systems, what portion of the buildout would be tied to training versus inference, or what timelines would apply.

Nor does the report provide details on how the proposed $45 billion figure would map to physical data centers, power and cooling capacity, or contracted service levels. Those elements are typically central in large AI infrastructure projects, but they were not spelled out in the market article.

Still, the reported size of the plan underscores how quickly AI infrastructure requirements can become intertwined with corporate financing decisions. Even when AI progress is measured in models and research, the bottleneck for many teams is frequently “how much compute can you reliably run,” which in turn is shaped by data-center capacity and procurement cycles.

Investors and industry watchers will likely monitor whether Anthropic confirms the talks, provides clearer deal terms, or updates any IPO-related timelines. In the near term, Nvidia’s attention will be on evidence of incremental infrastructure demand emerging from large model developers, particularly those scaling beyond current deployments.