#AnthropicSigns35BCloudDeal

Anthropic’s reported $35B cloud deal shows where the AI race is heading next: computing power is becoming just as important as the models themselves.

The headline is huge, but the deeper story is the infrastructure behind AI. Anthropic needs enormous amounts of compute to train and operate increasingly capable models, and securing large-scale cloud capacity can give an AI company more predictable access to the resources required for expansion.

The reported agreement with Lambda is especially interesting because it highlights the growing role of specialized cloud providers in the AI ecosystem. Instead of relying on a single traditional hyperscaler for every workload, leading AI companies are increasingly looking across the market for additional capacity, flexibility and access to advanced hardware.

Nvidia’s reported involvement through credit backing adds another layer. Nvidia is already central to the AI infrastructure stack because its accelerators power a large portion of modern AI workloads. Supporting financing around major compute commitments shows how the AI hardware ecosystem can extend beyond simply selling chips. Capital, hardware, networking and cloud capacity are increasingly becoming connected pieces of the same growth cycle.

This is why I think the real competition is no longer only Anthropic vs OpenAI vs Google. There is now an entire infrastructure race happening underneath the model race. Chip manufacturers, cloud providers, data-center operators, networking companies and energy suppliers all have a role in determining how quickly AI companies can scale.

The $35B figure also puts the economics of frontier AI into perspective. Building powerful models is extremely capital-intensive. Training is expensive, but serving millions of users can create another enormous and recurring demand for computing resources. As AI adoption grows, companies need infrastructure that can scale with that demand without becoming a bottleneck.

There is also a strategic advantage to securing capacity early. If demand for AI compute continues increasing faster than new data-center capacity can be built, access to chips and cloud infrastructure could become a competitive advantage. Companies with strong capital positions and long-term capacity agreements may be better positioned when the market becomes supply-constrained.

But there is another side to the story: $35B is an enormous commitment. The AI industry is moving at incredible speed, and today's assumptions about model demand, hardware efficiency and revenue growth may look very different a few years from now. Massive infrastructure spending can create an advantage, but it also increases the pressure to turn compute investment into sustainable revenue.

For Nvidia, the development reinforces how deeply the company is embedded in the AI expansion cycle. The opportunity is enormous, but so are the expectations. If AI infrastructure spending keeps accelerating, Nvidia remains strategically important. If customers eventually demand lower costs or become more efficient with compute, the economics of the ecosystem could change.

For Lambda, deals of this scale can represent a major opportunity to establish itself as an important AI infrastructure provider. The competitive landscape, however, remains intense. Traditional cloud giants and specialized GPU clouds are all fighting for a share of the same rapidly expanding market.

My bigger takeaway is simple: the next phase of the AI race will be fought in data centers. Better models matter, but models need chips, electricity, networking, storage and reliable computing capacity to reach users at scale.

The companies controlling that infrastructure could become just as strategically important as the companies building the AI applications themselves. The AI arms race is no longer just about who creates the smartest model. It is increasingly about who can secure enough compute to keep scaling.

$ANTHROPIC ‌