Elon Musk’s xAI is not letting regulatory hurdles or community protests slow down its quest for artificial general intelligence (AGI). In a move that highlights the desperate, high-stakes scramble for power in the artificial intelligence sector, xAI is committing a staggering $2.8 billion to purchase natural gas turbines over the next three years.

This massive capital expenditure was quietly revealed in a recent initial public offering (IPO) filing for SpaceX, another corner of Musk’s sprawling industrial empire. The disclosure sheds light on how Musk is leveraging his broader corporate ecosystem to solve the most pressing bottleneck in modern AI development: electricity.

This multibillion-dollar energy play comes at a highly sensitive time. xAI is currently embroiled in a high-profile legal battle over its "Colossus" supercomputer cluster in Memphis, Tennessee. Local community groups and environmental organizations, represented by the Southern Environmental Law Center (SELC), have sounded the alarm over the facility’s heavy reliance on unpermitted, polluting generators.

By locking in a massive fleet of natural gas turbines, Musk is signaling that xAI intends to bypass traditional utility timelines and generate its own power on an unprecedented scale, regardless of local resistance and legal challenges.

The revelation of the $2.8 billion purchase agreement in a SpaceX IPO filing underscores the deeply intertwined nature of Musk’s companies. While SpaceX and xAI are legally distinct entities, they frequently share resources, talent, and infrastructure. The filing indicates that the procurement of these natural gas turbines will span the next three years, ensuring a steady stream of heavy power-generation hardware to fuel xAI's rapid expansion.

This cross-company arrangement allows xAI to leverage the manufacturing, logistics, and purchasing power of SpaceX to secure critical energy infrastructure that would otherwise take years for a startup to acquire.

To understand the scale of xAI's energy hunger, one must look at the Colossus supercomputer. Touted by Musk as the most powerful AI training cluster in the world, Colossus utilizes 100,000 liquid-cooled Nvidia H100 GPUs, with plans to double that capacity to include Nvidia’s next-generation Blackwell chips.

Running a cluster of this magnitude requires an immense amount of electricity—estimated at up to 150 megawatts (MW) for the current setup, and potentially climbing to 300 MW or more once expanded.

When xAI set up shop in Memphis, the local utility, Memphis Light, Gas and Water (MLGW), and the Tennessee Valley Authority (TVA) could not immediately guarantee the massive amounts of power required. Rather than waiting years for grid upgrades, Musk opted for a "speedrun" approach, installing dozens of combustion turbine generators to bridge the gap and get the supercomputer online in record time.

This DIY power plant approach quickly drew the ire of Memphis residents and environmental watchdogs. Critics point out that burning massive quantities of fossil fuels in a densely populated area contributes significantly to smog, nitrogen oxide emissions, and local health hazards. The Shelby County Health Department and federal regulators have faced intense pressure to intervene, leading to formal complaints and lawsuits alleging that xAI operated these generators without the proper Clean Air Act permits.

The transition to $2.8 billion worth of dedicated natural gas turbines suggests that xAI is moving from temporary, stop-gap generator solutions to a permanent, captive natural gas-powered microgrid. While natural gas burns cleaner than diesel, it remains a fossil fuel that releases significant greenhouse gases, drawing sharp criticism from environmentalists who argue that the AI boom is actively derailing global climate goals.

xAI’s aggressive energy strategy is a symptom of a broader crisis facing the entire artificial intelligence industry. As tech giants race to train larger large language models (LLMs) and deploy agentic AI systems, the demand for electricity is outstripping grid capacity worldwide.

Other tech giants are pursuing alternative, albeit equally ambitious, energy strategies:

  • Microsoft has signed a massive deal to help resurrect a unit of the Three Mile Island nuclear plant.
  • Amazon Web Services (AWS) recently purchased a data center campus directly connected to a nuclear power plant in Pennsylvania.
  • Google is exploring geothermal and advanced nuclear technologies to power its future facilities.

Musk’s approach, however, is uniquely aggressive and immediate. By investing billions directly into natural gas infrastructure, xAI is prioritizing speed-to-market above all else. In the race for AGI, a delay of six months to wait for a clean utility connection could mean falling irreparably behind competitors like OpenAI and Google.

As xAI prepares to train its next-generation models, the $2.8 billion turbine purchase ensures that Colossus will have the power it needs to keep spinning, even if it means fighting off regulators and local communities in court. For the AI industry, it is a stark reminder that the virtual frontier of intelligence remains deeply bound to the physical, resource-constrained realities of the material world.