SpaceX is pouring roughly $2.8 billion into natural-gas turbines over the next three years, all to keep the lights on at its expanding fleet of AI data centers. The commitment, revealed in an IPO filing dated May 20, 2026, underscores just how desperate the AI industry has become for reliable power, even if that power comes with a thick cloud of regulatory and environmental baggage.

About $2 billion of that figure is earmarked for mobile gas turbines. These are the same units that have already landed xAI, SpaceX’s AI division, in legal and regulatory hot water near Memphis, Tennessee.

The power problem nobody wants to talk about

S&P Global estimates that data centers will add 11.3 gigawatts of power demand in 2025 alone. The International Energy Agency projects that AI-driven data center growth could require an additional 945 to 1,000 terawatt-hours of electricity by 2030, roughly the entire annual electricity consumption of Japan, created from scratch in under five years.

Grid interconnection delays can stretch years. Mobile gas turbines let xAI generate its own power on-site, no grid connection required. OpenAI, Microsoft, and Amazon are all exploring on-site power generation to keep their AI workloads humming without waiting in a multi-year queue for grid access.

Environmental pushback is already here

The NAACP has filed a lawsuit against xAI over allegedly unpermitted mobile gas turbines operating near Memphis, Tennessee. The complaint frames the issue as both an environmental and civil-rights matter, arguing that the pollution burden falls disproportionately on nearby communities.

Each turbine reportedly emits over 2,000 tons of nitrogen oxides (NOx) annually. NOx is a key ingredient in smog and is linked to respiratory disease.

The EPA has found xAI to be in violation of federal environmental laws regarding these generators this year.

What this means for crypto and the broader compute economy

The IEA’s projection of up to 1,000 TWh in new data center demand by 2030 would fundamentally reshape energy markets globally. Natural gas prices, carbon credit markets, and the economics of renewable energy buildouts are all affected when companies with multi-billion-dollar budgets start buying turbines at scale.

For Bitcoin miners and proof-of-work networks, the implications are direct. Higher energy demand from AI tends to push electricity prices up across the board. Miners operating on thin margins in deregulated markets, particularly in Texas and the US Southeast, could see their profitability eroded as AI-driven demand tightens local power supplies.

If the EPA’s enforcement actions against xAI lead to stricter permitting requirements for on-site gas generation, those same rules would likely apply to any compute facility, crypto or AI, trying to run its own power. The NAACP lawsuit could set precedents that ripple far beyond Memphis.