The race for compute capacity is no longer just a sprint; it’s a full-blown war of attrition. And Anthropic, the team behind the highly regarded Claude models, just signaled they are ready to play on the biggest field. Reports confirm the AI safety-focused firm has entered into a staggering $10 billion agreement with Volta, a relatively new cloud startup.
This isn’t your typical vendor contract. It is a strategic fortification.
Volta, which launched earlier this year, is set to supply cloud processing power to Anthropic for the next six years. Six years. In AI terms, that is an eternity. It suggests Anthropic isn’t just looking to solve immediate latency issues; they are planning for a massive scale-up that likely spans multiple generations of hardware.
But here is the twist: Volta isn’t doing this alone. They are bringing in Bitdeer, a company primarily known for crypto mining, to help build out the infrastructure. Yes, you read that right. Crypto miners pivoting to power generative AI giants. The synergy is oddly logical—both sectors require dense, high-efficiency compute resources—and it highlights how fluid the lines have become in the tech landscape.
Where the Power Comes From
The physical location matters as much as the money. This specific deal involves a new data center in Norway. Why there? Likely because of renewable energy availability and stable geopolitical footing. The facility will boast a 133-megawatt capacity. For context, that is enough electricity to power tens of thousands of homes, yet it will be entirely dedicated to training and running large language models.
And what will those servers run? Nvidia’s latest architecture. The setup is fueled by Nvidia’s Vera Rubin systems. Vera Rubin represents the chipmaker’s state-of-the-art AI chip design. This confirms that despite the hype around domestic silicon or custom chips from companies like Groq or Cerebras, the industry standard remains entrenched in Santa Clara.
Volta’s inclusion in Nvidia’s Cloud Partner program—a consortium of providers using Nvidia GPUs in their data centers—validates the deal’s technical backbone. It’s not a fringe experiment. It’s the mainstream path, just scaled up to a level most companies can’t comprehend.
Why This Matters for the AI Market
You might ask why a $10 billion number should make you pause. It’s because capacity is the new oil. Or rather, it’s
