News
Elon Musk Quietly Bought a Gas Turbine Company for $1 Billion to Power AI

Elon Musk acquired APR Energy, formerly owned by Fortress Investment Group, for roughly $1 billion in a deal that only surfaced through FTC filings. The mobile gas and diesel turbines are meant to power xAI's power-hungry data centers, including the Colossus complex in Memphis.
For several months, Elon Musk secretly bought an energy company meant to solve his biggest obstacle to scaling artificial intelligence: a lack of electricity. The roughly $1 billion deal only came to light through filings with the US Federal Trade Commission (FTC), not through any official announcement.
APR Energy is a Jacksonville, Florida-based company known for supplying temporary power plants on wheels. Its gas turbines and diesel engines are mounted on trailers, letting them go online in days rather than the years needed to build a traditional power plant and connect it to the grid. That flexibility is exactly what appeals to Musk, whose companies have spent months struggling with power shortages while scaling their AI infrastructure.
Why power, not chips
For a while now, the biggest constraint on further AI model development hasn't been graphics chips, it's access to electricity. US power grids can't keep up with connecting new, gigantic data centers, and getting approval to build a permanent power plant can take years. APR Energy's mobile turbines sidestep that problem, since they can be set up right next to a server hall almost immediately.
The capacity from the acquired company is set to go mainly toward xAI's infrastructure, Musk's company behind the Grok model. The key recipient is the Colossus computing complex in Memphis, Tennessee, which has been expanding for months to support the next generations of training supercomputers.
The shadow of a pollution lawsuit
The APR Energy purchase isn't a neutral topic for xAI. Musk's company has already faced a lawsuit over using similar mobile gas turbines without the required permits at a data center in Southaven, Mississippi, just across the border from Memphis. The allegations concerned violations of the federal Clean Air Act, and the number of turbines installed there grew at a pace that alarmed local regulators and environmental groups.
Acquiring its own turbine supplier gives xAI greater control over where and under what terms it sources equipment, but it doesn't resolve the underlying tension between rapidly scaling computing power and environmental requirements. FTC filings also show the agency expedited its antitrust review of the deal, waiving part of the standard waiting period.
Musk versus the power grid
This is another move in Musk's broader strategy of building his own power supply for his AI empire instead of waiting for public energy infrastructure to catch up. SpaceX has already committed to spending more than $2.8 billion on gas turbines for data centers, and the APR Energy acquisition means part of that money will now go to a company Musk controls directly rather than to an outside supplier.
Notably, according to FTC filings, the buyer listed in the transaction is Elon Musk personally, not one of his companies outright. It's an unusual structure that makes it harder to track the flow of assets between his businesses and may explain why the purchase wasn't made public earlier.
For energy market observers, this signals that AI power infrastructure is becoming its own, highly valued investment segment. Companies like APR Energy, previously associated mainly with emergency power after natural disasters or in developing countries, are suddenly becoming a critical link in the supply chain for the biggest AI players.
For Polish readers, the case illustrates the scale at which energy costs are starting to define the global AI race. While in Poland the debate over AI data centers mostly concerns resident protests and strain on local grids, in the US the biggest players are going a step further and buying entire energy companies to secure their own power supply regardless of the state of public infrastructure.
The case is also a reminder that developing models like Grok, or competing systems from OpenAI and Google DeepMind, has a very physical, energy-intensive dimension that's becoming harder to hide behind announcements of the next software release.

