Anthropic, the developer of the Claude models, has signed a six-year, $10 billion deal with Volta Infrastructure Holdings. This is not a traditional equity investment in a start-up, but rather the purchase of the infrastructure and computing power required to develop and operate AI models. Volta was only founded in January 2026, yet its valuation already stands at around $2.4 billion. Investors include Nvidia and Michael Dell’s family fund, amongst others.
Volta’s first major project is set to be a data centre in Norway utilising Nvidia’s Vera Rubin systems. The infrastructure is linked to Bitdeer, a company with its roots in bitcoin mining, which is increasingly shifting its focus towards AI. Bitdeer had previously announced plans to expand its Norwegian Tydal facility to 180 MW of capacity.
The deal demonstrates that in the AI race, it is no longer just models and data that give a competitive edge, but above all access to energy, chips and data centres. In May, Anthropic secured $65 billion in funding at a valuation of $965 billion, explicitly stating that part of the funds would be allocated to increasing available computing power.
This is not an isolated case. In July, Anthropic also signed a 20-year contract with TeraWulf for a data centre with a capacity of around 401 MW, valued by the operator at approximately $19 billion.
This could lead to a further rise in the value of companies with access to energy and existing infrastructure. At the same time, long-term contracts increase the financial risk for AI laboratories: today’s shortage of computing power may not last forever, whilst the commitments will remain.
