Anthropic has secured one of the largest AI infrastructure commitments in the industry’s history. On August 11, 2026, Riot Platforms disclosed a $9.1 billion, 20-year data center lease at its Rockdale, Texas campus — with Bloomberg reporting Anthropic as the counterparty, citing people familiar with the deal.

The numbers are significant. Neither company confirmed the arrangement publicly by name, but the market responded immediately: Riot Platforms stock surged 17% on the news.

The Deal Details

The agreement covers 191 megawatts of IT capacity at Riot’s Rockdale facility, with a lease term running through June 2048. Riot plans to bring 96 megawatts online by December 2027 and complete the full 191-megawatt buildout by June 2028.

Two five-year extension options are embedded in the contract, which could push the total contract value to $16.1 billion if exercised.

Riot originally disclosed the deal without naming its counterparty, describing them only as a “leading frontier AI lab.” That description matched Anthropic, and Bloomberg confirmed the identity through sources familiar with the matter.

For Anthropic, this follows a series of large compute agreements it has been pursuing as demand for its products — particularly Claude Code and Claude’s enterprise APIs — has accelerated. The company has been pursuing large computing agreements as Claude usage scales across enterprise customers.

Why This Matters: The Compute Race Is Structural

What makes this deal notable is not just the scale — it is the term. A 20-year lease is a structural bet, not a tactical one. Anthropic is making a long-horizon commitment to Texas-based compute infrastructure, locking in capacity through 2048.

The AI industry has hit a consistent constraint: demand for inference compute consistently outpaces what cloud providers can deploy. Companies at the frontier — OpenAI, Anthropic, Google DeepMind — are increasingly going direct to power and real estate, bypassing traditional cloud pricing entirely.

Riot Platforms, originally a Bitcoin mining company, converted a portion of its Rockdale site from mining operations to AI data center capacity. That conversion story — from proof-of-work compute to AI inference compute — reflects a broader shift in how and where AI capacity gets built.

What This Means for Enterprise Claude Users

In practical terms, this deal signals that Anthropic is building toward sustained availability at scale. For enterprise teams that have been hitting capacity limits or experiencing rate throttling on Claude’s API, this kind of infrastructure investment is what resolves those constraints over the medium term.

More capacity means:

• Better API availability for teams running high-volume workloads — automated reporting, document processing, multi-agent pipelines
• Increased competition on inference pricing as Anthropic’s owned compute costs decline relative to rented cloud infrastructure
• More room for long-context workloads — the memory-intensive tasks that strain shared infrastructure fastest

The timeline matters here. With 96 megawatts coming online by late 2027 and full capacity in 2028, this is not an overnight fix — but it is a credible buildout plan for teams thinking about where to anchor their AI infrastructure stack in the next three to five years.

The Bigger Infrastructure Story

Anthropic is not alone in this pattern. Amazon, Google, and Microsoft have all made multi-billion-dollar commitments to AI data centers this year. The difference with this deal is the counterparty: Riot Platforms is not a traditional cloud vendor. It is a converted crypto miner with large power contracts in Texas — exactly the kind of asset that becomes valuable when electricity and real estate are the constraints that matter.

The AI infrastructure buildout increasingly looks less like software scaling and more like power utility planning. Companies that secure long-term compute capacity at fixed pricing have a structural cost advantage over those paying spot rates on shared cloud.

For enterprise buyers, the lesson is familiar: the AI vendors making durable infrastructure investments today are the ones most likely to offer stable, scalable access tomorrow.

---

What This Means for Business: Anthropic is securing the long-term compute it needs to scale Claude reliably. For enterprise teams building AI-dependent workflows, this reduces infrastructure risk when choosing Anthropic as a core AI platform.

Enterprise DNA builds AI-powered operations and data workflows for mid-market and enterprise teams. Explore Omni by Enterprise DNA to see how we design AI systems that scale.