Elon Musk is working to resolve GE Vernova's AI power bottleneck.

Elon Musk is attempting to manufacture his way around one of the AI boom’s most pressing constraints: securing sufficient electricity to power the industry’s expensive Nvidia Corp. (NASDAQ:NVDA) chips. To address this bottleneck, SpaceX (NASDAQ:SPCX) is constructing a foundry in Bastrop, Texas, designed to cast blades and vanes for industrial gas turbines. According to The Information, which cited recent SpaceX job postings detailing the facility’s construction and operational ramp-up, the project represents a direct push into heavy manufacturing. Musk subsequently confirmed on X that SpaceX is bringing the casting process in-house, noting the move could accelerate the deployment of natural-gas turbines by up to 18 months.

This strategic pivot directly intersects with a supply chain segment GE Vernova Inc. (NYSE:GEV) has repeatedly flagged to investors. GE Vernova CEO Scott Strazik has stated that the company has been actively pressuring suppliers to expand capacity merely to match surging turbine demand. At a Bernstein conference in May, Strazik revealed that GE approached its castings and forgings suppliers during the summer of 2024, demanding substantially higher output. In certain instances, GE even financed the construction of new furnaces for those suppliers. That pressure has intensified since then. GE remains largely sold out through 2030 and is working to scale annual turbine production from approximately 20 gigawatts this year to 30 GW by the decade’s end.

Achieving that production ramp hinges precisely on the component supply chain Musk is now moving vertically. During GE’s July earnings call, Strazik emphasized that reaching 24 GW of annual output in 2028 is contingent upon receiving additional castings and forgings throughout 2027. Musk had previously signaled this necessity in February, stating that both SpaceX and Tesla may need to manufacture blades and vanes internally due to severe casting capacity constraints. SpaceX has already allocated more than $2.8 billion toward gas turbines over a three-year period to power its Colossus data centers near Memphis. Developing blade and vane production in-house would extend that investment further upstream, directly targeting the supply chain friction that currently dictates turbine delivery timelines.

The urgency is underscored by rapid data center expansion. Data Center Map currently tracks 4,767 U.S. data centers, while Kalshi traders assign a 66% probability that the count will reach at least 5,300 by year-end. This trajectory implies hundreds of new facilities entering the grid queue over the coming months, intensifying competition for available power. Accelerated turbine component manufacturing would alleviate only a portion of the bottleneck. SpaceX must still demonstrate it can produce these parts reliably at commercial scale, and as Strazik has noted, broader project timelines remain vulnerable to construction delays, permitting hurdles, fuel supply logistics, and ancillary infrastructure gaps.

For years, the central challenge in artificial intelligence infrastructure was securing advanced semiconductors. The focus has now decisively shifted toward securing the electrical capacity required to operate them.

Kalshi and Benzinga operate under an existing data collaboration agreement. Separately, investor Steve Eisman has warned that a potential failure by OpenAI could trigger an immediate recession and a broad technology sector selloff, even as current market pricing reflects only an 8% probability of a 2026 recession.