In a move that underscores the immense energy demands of next-generation artificial intelligence, Elon Musk has confirmed that his AI startup, xAI, has purchased five massive natural gas turbines from South Korean energy heavyweight Doosan Enerbility. The acquisition, aimed at powering xAI’s rapidly expanding supercomputer clusters, marks a significant step in the company’s aggressive strategy to secure independent, reliable energy infrastructure for its high-performance computing needs.
The confirmation came via Musk’s social media platform, X, following reports that detailed the supply of heavy-duty power generation equipment to a major U.S. tech entity. By securing nearly 2 gigawatts of power generation capacity independently, xAI is positioning itself to bypass traditional utility constraints, ensuring that its training of future models—such as the anticipated Grok 5—proceeds without the bottlenecks that have plagued other players in the AI sector.
This development follows closely on the heels of xAI’s massive financial injection, with the company recently closing an upsized Series E funding round. With billions in fresh capital and a clear roadmap for infrastructure scaling, the purchase of these turbines signals a transition from software development to industrial-scale physical infrastructure management, setting a new precedent for how AI companies may operate in the future.
Confirming the Power Play
The revelation regarding the turbine purchase began with a report shared by user @SemiAnalysis_ on X, which identified the equipment supplier as Doosan Enerbility. The analysis connected the dots between Doosan’s recent contract announcements and xAI’s known infrastructure projects. According to the Asian Business Daily, Doosan Enerbility had announced a contract signing last October to supply two 380 MW gas turbines to a major U.S. technology company. This was followed by a subsequent announcement in December regarding an order for three additional turbines of the same specification.
When the report circulated on X, speculating that these units were destined for xAI’s facilities, Elon Musk provided a typically concise verification. Replying to the thread detailing the purchase and its implications for xAI’s cluster size, Musk simply wrote:
“True”
This single-word confirmation validated months of speculation regarding how xAI intended to power its “Colossus” supercomputer and future expansions. The five turbines, each capable of generating 380 megawatts, represent a combined potential output of 1,900 megawatts—or 1.9 gigawatts. To put this figure into perspective, this amount of power is comparable to the output of two standard nuclear reactor units, highlighting the sheer scale of energy required to train frontier AI models.
The Hardware Behind the Heat: Doosan Enerbility’s Role
The choice of Doosan Enerbility as the supplier highlights the specialized nature of the equipment required for this undertaking. The South Korean firm is a global leader in power plant facilities and desalination plants, known for producing robust, high-efficiency gas turbines. The 380 MW class turbines are industrial behemoths designed to provide baseload power, typically used by utility companies to power hundreds of thousands of homes.
By purchasing these turbines directly, xAI is effectively becoming its own utility provider. In the current landscape of AI development, access to power has become a more significant constraint than access to silicon. While competitors struggle with long lead times for grid connections and utility upgrades, xAI’s strategy to install on-site generation allows for rapid deployment. This aligns with Musk’s historical preference for vertical integration, seen clearly in Tesla’s supply chain management and SpaceX’s in-house manufacturing.
The deployment of these five turbines suggests a phased expansion. With the initial order of two units followed by three more, xAI appears to be scaling its power infrastructure in tandem with the delivery of its compute hardware, ensuring that as new server racks arrive, the electricity needed to run them is already available.
Fueling the Compute Engine: The GB200 NVL72 Cluster
The primary purpose of this massive energy procurement is to fuel what promises to be one of the world’s largest computing clusters. According to the analysis confirmed by Musk, the gas turbines are intended to power an additional cluster equivalent in size to 600,000+ GB200 NVL72 units. The GB200 NVL72 is part of Nvidia’s Blackwell architecture, a rack-scale design that integrates 72 Blackwell GPUs and 36 Grace CPUs into a single, liquid-cooled unit.
This architecture represents a quantum leap in compute density and efficiency but comes with substantial power requirements. A single rack of NVL72 can consume upwards of 120 kilowatts. Scaling this to the numbers suggested by the analysis requires not just a connection to the grid, but a dedicated power plant. The 1.9 gigawatts of capacity provided by the Doosan turbines will be essential to maintain the thermal and electrical stability of such a dense computing environment.
This infrastructure buildout is critical for xAI’s mission to catch up to and surpass its rivals. The company has stated that its facilities, once fully operational with this new hardware, will be among the largest in the world. This capacity is not merely for bragging rights; it is a prerequisite for training the next generation of large language models (LLMs), which require processing datasets of unfathomable size.
Financial Firepower: The Series E Funding
Such ambitious infrastructure projects require immense capital, and xAI has successfully secured the necessary funds. Recent reports indicate that the company closed an upsized Series E funding round, raising $20 billion. This figure significantly exceeded the initial target of $15 billion, reflecting strong investor confidence in Musk’s vision and the company’s trajectory.
The company has been transparent about the intended use of these funds. In a statement regarding the financing, xAI noted that the capital would be used to “accelerate our world-leading infrastructure buildout, enable the rapid development and deployment of transformative AI products.” The purchase of the Doosan turbines is a direct execution of this mandate. By front-loading the investment in power generation, xAI ensures that its capital expenditure translates directly into operational velocity.
The oversubscription of the Series E round suggests that the market views xAI not just as a software startup, but as an infrastructure play. Investors are betting that the company which controls the most compute and power will ultimately dominate the AI landscape. With a valuation that has soared in light of these developments, xAI is utilizing its war chest to create a formidable moat around its operations.
Looking Ahead: Grok 5 and Beyond
The ultimate beneficiary of this power and silicon is xAI’s product line, specifically the Grok series of AI models. The company has teased the development of its next frontier model, Grok 5, stating that it is currently in training. The computational intensity required to train a model like Grok 5—which is expected to reason, code, and understand the world with superhuman proficiency—is the driving force behind the turbine acquisition.
On its website, xAI outlined its forward-looking strategy:
“Looking ahead, Grok 5 is currently in training, and we are focused on launching innovative new consumer and enterprise products that harness the power of Grok, Colossus, and 𝕏 to transform how we live, work, and play.”
The mention of “Colossus” refers to the supercomputer cluster located in Memphis, Tennessee, which has already broken records for the speed of its deployment. The addition of the new turbines suggests that Colossus is set to grow significantly, or that new, even larger clusters are being established. The integration with 𝕏 (formerly Twitter) provides xAI with a unique distribution channel and a real-time data source, creating a feedback loop that accelerates model improvement.
The Strategic Imperative of Energy Independence
The move to purchase natural gas turbines also highlights a critical bottleneck in the AI industry: the electrical grid. Across the United States, data center demand is outpacing the utility sector’s ability to build new transmission lines and generation capacity. By opting for on-site natural gas generation, xAI is decoupling its growth from the slow-moving regulatory and construction timelines of public utilities.
While natural gas is a fossil fuel, modern turbines like those from Doosan are highly efficient. Furthermore, the stability provided by gas turbines is crucial for AI training runs, which can last for months. A power interruption of even a few milliseconds can disrupt a training checkpoint, potentially costing millions of dollars and weeks of lost time. Having dedicated, on-site power generation offers a level of reliability that the commercial grid often cannot guarantee for loads of this magnitude.
This strategy effectively allows xAI to turn money into intelligence at a rate limited only by the delivery speed of turbines and GPUs, rather than the permitting speed of local governments. It is a brute-force approach to innovation that characterizes much of Musk’s career, prioritizing speed and scale above all else.
Conclusion
Elon Musk’s confirmation of the purchase of five 380 MW natural gas turbines serves as a stark reminder of the physical realities underpinning the artificial intelligence revolution. As xAI races to train Grok 5 and expand its Colossus cluster, the convergence of heavy industry and advanced software becomes increasingly apparent. With $20 billion in fresh funding and nearly 2 gigawatts of independent power generation secured, xAI is aggressively clearing the path toward artificial general intelligence.
The deal with Doosan Enerbility is more than a procurement contract; it is a declaration of intent. xAI is building the industrial machinery necessary to support the next era of computing, ensuring that when the next breakthrough in AI architecture arrives, they have the power—literally and figuratively—to seize it.
