Shares of turbine blade and vane manufacturers took a hit, yet the competitive threat posed by the tech mogul may be more theoretical than immediate.

Only a select group of companies can produce the essential gas turbine parts that AI data centers are increasingly depending on. Now, Elon Musk has indicated plans to manufacture these components himself, causing unease among investors in this tightly controlled market.

Over the weekend, Mr. Musk confirmed on his social media platform that SpaceX would begin in-house production of blades and vanes for natural gas turbines. He stated that by casting parts internally, they could accelerate the deployment of gas turbines by as much as 18 months, calling it a potential game-changer. This came after a report from Morgan Stanley, which had earlier flagged SpaceX's intention to build a foundry in Bastrop, Texas, based on its job postings.

News of the plan sent Howmet Aerospace Inc. shares down more than 7% on Monday, with only a partial recovery seen on Tuesday. Howmet Aerospace Inc. and Berkshire Hathaway-owned Precision Castparts are the dominant players in this niche, supplying blades and vanes not just for power generation but also for aerospace and defense applications. Smaller rival DPC also saw its stock price dip on the announcement.

It may be premature to view SpaceX as a serious competitor right now. However, the recent pullback in shares of Howmet Aerospace Inc. and DPC could be viewed as a correction, given their rich valuations. They were trading at forward price-to-earnings ratios of 42 and 45, respectively, making them some of the most expensive picks in the AI infrastructure space. In comparison, turbine manufacturers GE Vernova and Caterpillar trade at multiples of 33 and 26.

Among the smaller players, DPC appears most exposed. It generates roughly 40% of its revenue from gas turbines and has a smaller scale. Howmet Aerospace Inc., while holding about half of the global market for gas turbine components, is more diversified, with only 11% of its revenue coming from that segment.

Entering this field isn't straightforward. The technical challenges are immense; blades must endure extreme heat and high-speed rotation, requiring a complex manufacturing process. Each part is created from a unique wax pattern, and the casting must be performed in a vacuum furnace to form a single-crystal nickel-based superalloy. Even minuscule flaws, like stray grains or hairline cracks, can render a component unusable. Industry analysts suggest that a new production line could see defect rates exceeding 50% for a considerable time.

Adding to the difficulty is the high level of secrecy surrounding the process. Howmet Aerospace Inc. executives have noted that ceramic cores and wax molds are destroyed after production to prevent reverse engineering by competitors.

If SpaceX produces these parts for its own use, high scrap rates may be less of a concern. Morgan Stanley's report suggests it's unlikely the company will become a major external supplier. Having an in-house source for replacement parts could be convenient for Musk's xAI data centers, especially considering the fluctuating power demands of AI workloads. SpaceX is already familiar with advanced engine technology, designing and manufacturing many of its own rocket engine and avionics components.

The proposed foundry is expected to produce single-crystal nickel-based superalloy parts for the turbine pumps of its Raptor rocket engines as well as for gas turbines. It remains unclear if SpaceX has already begun casting its own blades for the Raptor pumps. Even if it has, industry experts estimate a new facility could take at least four years to become operational—a long time in the fast-paced world of AI. The lead time for specialized vacuum furnaces can exceed two years, and it takes considerable time to hire and train a skilled workforce. According to DPC, key positions require six to twelve months of training, and their technical staff often stay for nearly a decade.

If anything, Musk's foundry plan underscores the critical nature of these blades and vanes for AI infrastructure. The recent drop in share prices gives investors a chance to evaluate what they are willing to pay for such market power, but it doesn't fundamentally undermine the investment thesis for Howmet Aerospace Inc. or DPC.