xAI Secures Major Power Deal with Doosan Enerbility to Fuel AI Expansion

xAI Secures Major Power Deal with Doosan Enerbility to Fuel AI Expansion

January 6, 2026

In a significant move highlighting the intensifying energy demands of artificial intelligence development, Elon Musk's xAI has placed a major order for natural gas turbines to power its next-generation computing clusters. The deal underscores a critical trend in the industry: the race for AI supremacy is increasingly becoming a race for reliable, high-capacity power.

The company has purchased five natural gas turbine units from South Korea's Doosan Enerbility, with a total capacity of 380 megawatts (MW) per unit. According to industry analysis firm SemiAnalysis, which first reported the deal, the first two turbines are scheduled for delivery by the end of 2026. This substantial power infrastructure is earmarked to support a massive new AI computing cluster.

The newly secured power will be dedicated to running what SemiAnalysis estimates to be the equivalent of over 600,000 Nvidia GB200 NVL72 systems. Alternatively, the same energy could power a cluster equivalent to more than 350,000 of the even larger VR200 NVL144 systems. This scale of procurement places xAI's planned data center footprint among the largest and most power-intensive in the world dedicated to AI training.

Commenting on the strategic implications, SemiAnalysis noted on social media platform X that xAI, under Musk's leadership, is "singular in terms of their ability to build the absolute largest data center in the world," adding that "nobody else can run as fast as them." This move positions xAI to secure the foundational energy resources required for training increasingly complex large language models and other advanced AI systems, a process known for its enormous electricity consumption.

The deal signals a strategic shift for AI companies toward directly managing their power destiny, often bypassing strained regional grids. By investing in on-site or dedicated power generation, firms like xAI aim to guarantee the stability and scale of electricity needed for continuous, uninterrupted operation of their AI superclusters. This trend is likely to accelerate competition not just in chip technology, but also in securing energy assets and partnerships with heavy industrial equipment manufacturers like Doosan Enerbility.

Source: SemiAnalysis

Read Also
Lubbock City Council to Decide Fate of Proposed AI-Powered Hyperscale Data Center
McDuffie County Considers Proposal for Major New Data Center Development
Data center issuers get cracking on bumper year with $1.45bn of deals

Research