Anthropic Locks Down $3.4bn Gas-Fired Power Deal as AI Data Centers Race for Energy

Anthropic's Massive Power Play

Anthropic has secured a $3.4bn deal for gas-fired power generators in what represents the largest known single purchase of power-generation equipment by an AI lab1

1

. The frontier AI company bought commitments for 1.7 gigawatts of heavy-duty generators in late September from OnePWR Solutions, a San Antonio, Texas-based equipment supplier2

2

. This transaction marks a significant shift in how AI companies are approaching the critical challenge of securing power for data centres as demand for electricity surges across the industry.

Bypassing Traditional Energy Suppliers

The Anthropic generator deal reveals a strategic move to cut out intermediaries in the race for scarce equipment. Rather than acquiring OnePWR itself, Anthropic took over the supplier's existing orders for Caterpillar generators that had been placed approximately two years earlier4

4

. OnePWR had ordered the generators but needed financing to complete the transaction, according to sources familiar with the matter1

1

. Anthropic originally planned to become a OnePWR customer but opted instead to take direct control of the equipment orders. The deal included engineering and construction capabilities needed to install the generators, though none of OnePWR's employees transferred to Anthropic3

3

.

AI Infrastructure Demands Drive Energy Sector Push

The surging power demands of AI are pushing tech companies deeper into the energy sector as manufacturing lead times for gas turbines and generators stretch years into the future1

1

. AI developers typically rely on utilities and specialist power companies to provide power for data centres, but the current market dynamics are forcing a fundamental shift in strategy. By securing access to generators ordered years earlier before prices surged, Anthropic has positioned itself to receive equipment in the near term and over the next two years1

1

. This approach mirrors moves by other tech giants—Elon Musk's SpaceX xAI bought mobile gas turbines earlier this year for $2bn, while Alphabet purchased energy developer Intersect in late 20251

1

.

Computing Power and Claude's Growing Demands

Strong demand for Anthropic's Claude family of models has prompted the company to secure additional computing power and capacity2

2

. The company revealed on Monday that it had nearly $900bn in computing commitments through 2045, having relied heavily on partners to secure power, chips, data centres and land while paying a premium to lock down capacity and meet demand1

1

. Buying generators directly gives Anthropic more control over the electricity needed for these facilities. Anthropic is also a major SpaceX customer, with monthly payments to lease computing capacity expected to reach $3.1bn by year-end1

1

.

Infrastructure Control and Market Implications

Arash Nazhad, managing director and head of clean energy and power at Stifel, noted that "by owning some of their power infrastructure and having the capabilities in-house, companies believe they can better protect themselves against disruptions beyond their control"1

1

. This strategic shift toward vertical integration in AI infrastructure reflects the industry's recognition that energy access has become a critical bottleneck. Anthropic is preparing for its stock market listing as early as this month in what could be one of the biggest listings in history, as it seeks to finance more than $895bn in computing infrastructure spending through 20451

1

4

4

. However, its ambitions face obstacles—Blackstone warned investors that it might receive a junk credit rating rather than the investment-grade previously expected1

1

. The company's amended IPO filing showed $895bn in commitments, almost all for computing power and infrastructure4

4

. Watch for how this deal influences other AI companies' energy strategies and whether direct equipment purchases become the new norm in an industry racing to secure the foundational resources needed to power next-generation AI systems.