IA · 5 August 2026 · 4 min read
Anthropic and the Compute Gold Rush: $10 Billion Deal with Volta Marks the Rise of Neoclouds
In brief: Anthropic has entered into a massive six-year agreement valued at $10 billion with AI cloud startup Volta to secure dedicated high-performance compute capacity powered by Nvidia's Vera Rubin architecture. Developed alongside Bitdeer in Norway, the project highlights how AI labs are increasingly turning to specialized neocloud infrastructure to feed their growing processing demands.
by Team Mocchi's
The Record $10 Billion Agreement Between Anthropic and Volta
The relentless demand for raw processing power to train and run frontier AI models continues to reshape the financial and infrastructure landscape of high tech. Anthropic, the creator of Claude, has entered into a massive six-year, $10 billion agreement with specialized AI cloud provider Volta. As reported by TechCrunch, the deal secures dedicated infrastructure in Norway providing 133 megawatts of capacity.
To build out the data center facility, Volta is partnering with Bitdeer, a crypto-mining operator that has repurposed its sites for high-performance enterprise computing. The Scandinavian facility will leverage Nvidia's latest Vera Rubin architecture, placing Volta firmly within Nvidia’s official Cloud Partner network.
The Rise of Neoclouds and Hardware Diversification
The multi-billion dollar transaction reflects a structural shift in the cloud market. Alongside traditional hyperscalers—such as Amazon Web Services, Google Cloud, and Microsoft Azure—a new ecosystem of specialized "neoclouds" is rapidly emerging. These providers excel at deploying specialized GPU clusters with optimized energy footprints and accelerated setup timelines.
For AI labs, diversifying hardware suppliers is not merely a reaction to ongoing chip shortages; it is a strategic effort to avoid lock-in to any single cloud ecosystem. Over recent months, Anthropic has pursued an aggressive infrastructure strategy, securing parallel compute agreements with established giants like Amazon alongside specialized players.
SpaceX and the New Boundaries of Compute Delivery
Illustrating how non-traditional players are entering the cloud market, even aerospace companies are becoming major providers of AI compute capacity. As highlighted by The Verge, SpaceX’s latest earnings reveal that its AI compute division generated $2.6 billion in revenue this quarter by renting out cluster capacity to third parties like Anthropic and Google—outpacing its revenue from space launches.
The conversion of former mining facilities, industrial sites, and private energy hubs into dedicated AI clusters is redrawing the global map of supercomputing. Proximity to reliable, green energy—such as Norway’s hydroelectric grid—has become the decisive factor in selecting next-generation data center locations.
Mocchi's take
The megadeal between Anthropic and Volta confirms that the race for AI supremacy is as much about physical compute and power distribution as it is about software architecture. For European companies building or deploying enterprise solutions on top of models like Claude, the expansion of high-efficiency compute clusters within European borders—such as the Norwegian deployment—is a highly positive signal for API latency, stability, and data sovereignty. At the same same, the sheer capital expenditure required to keep frontier models running reinforces why businesses must design vendor-neutral software architectures that can seamlessly shift workloads between models and cloud backends as inference costs fluctuate.