Howmet survives SpaceX: Bernstein's $328 bet on the 80% blade duopoly Musk is choosing to escape, not break

Harned's note draws a sharp line between SpaceX vertical-integrating its own turbine supply and SpaceX becoming a third-party casting supplier. \"Assuming SpaceX does the buildout to produce blades, we would not expect the company to become a new industry supplier of forgings and castings, which would have little connection with core SpaceX objectives,\" he wrote. \"We also are skeptical that SpaceX would be able to have volume production in time to meet its power expansion goals.\" The build targets20 GW of behind-the-meter capacity by end-2027, an internal Musk timeline, not a commercial sales channel.

The market disagreement hinges on whether SpaceX will stop at internal use. Citi sided with Bernstein the same week, calling the drop a buy-the-dip catalyst. The skeptical read — single-crystal directional solidification is one of the hardest metallurgical processes on Earth, with only a handful of qualified foundries worldwide — argues that even if SpaceX succeeds, it spends years solving the yield-and-coating stack rather than building merchant capacity. SpaceX itself outsources melt production of its proprietary SX alloys to an external mill under NDA, per industry reporting on its Raptor 3 turbopump program.

The high end of investment-cast turbine airfoils — single-crystal and directionally solidified nickel superalloys used in the hottest section of a gas turbine — is effectively a two-company market. Howmet Aerospace and Precision Castparts, owned by Berkshire Hathaway, hold roughly 80% combined share, with single-crystal concentrated even further. Doncasters (private, German-owned) and ATI cover the rest. A foundry cannot enter the market with just casting machines: it needs qualified alloys, vacuum-melt capacity, directional-solidification furnaces, single-crystal grain selectors, post-cast heat treatment, machining, performance coatings, hot isostatic pressing, and a decades-long aerospace qualification track record.

The technical moat is reinforced by the supply chain's deliberate underbuilding. Both Howmet Aerospace and Precision Castparts have been explicit about limited capex appetite for IGT blade capacity. The result is a market that has run short of supply even with no new entrant in sight, exactly the dynamic Bernstein points to. Musk's plan doesn't subtract from that shortage — at most, it removes one captive buyer from the queue.

Q2 2026 results, reported Aug 6, 2026, frame the gap between demand and supply. Total revenue of $2.55B was up 24% YoY with21% organic. Engine Products — where IGT blades and jet-engine airfoils live — grew third-party sales 32% to $1.37B, with adjusted EBITDA up 51% to $517M, a37.7% segment margin. Fastening Systems grew 37%. Industrial gas turbines as an end market grew 38%. The growth is being absorbed because Howmet Aerospace has been quietly adding capacity: a new Japan plant, a major European expansion, and a Virginia facility retooled by exiting some nickel-alloy work, with a new large casting machine going in and another following in the second half.

Capacity additions translate into38% more IGT blade output versus Q1 2025 by the end of 2026, on management's plan. Customer agreements run through 2030, and Howmet Aerospace has just finished negotiations with all seven of its major IGT customers, several of whom are already asking to revise demand outlooks higher. The bottleneck Musk is trying to escape is one HWM is actively widening.

SpaceX's foundry targets casting, not alloy production, so it would still buy nickel and cobalt superalloy billet, plate and powder from upstream melters. Carpenter Technology and ATI are the two listed US specialty-alloy producers exposed to that pull. Carpenter Technology, with ~60% of revenue from aerospace and defense and FY2026 revenue tracking above $2.8B, sells the nickel superalloys that go into both Howmet's blades and any new entrant's blades. ATI posted Q2 2026 sales of $1.26B, up 11% YoY, with jet-engine revenue up 13% and a record $4.4B backlog up 18% YoY. The takeaway: SpaceX's vertical move tightens alloy demand even if it leaves casting demand at Howmet Aerospace untouched.

RTX and GE Aerospace each represented about 11% of Howmet Aerospace's 2025 third-party sales, almost entirely from the Engine Products segment, per the 2025 10-K. Neither is at risk of losing supply in the next 24 months. The medium-term read is that HWM's 50%+ IGT share becomes a strategic chokepoint for the OEMs themselves, which is why the customer agreements run through 2030 and why HWM's biggest customer (Pratt & Whitney) is already building out a next-generation GTFA engine program with HWM content scaling into2027.

After the recovery, Howmet Aerospace trades at about $259 with a $103.7B market cap. EV/EBITDA on TTM sits around 39x and trailing P/E near 56x — rich on standard multiples, but in line with the broader aerospace supply-chain rerating since the GTF/Raptor ramp became visible in 2023-2024. Bernstein's $328 target implies ~27% upside; the Street consensus is $337 with a Moderate Buy rating skewed by 19 Buys vs 3 Holds. FY2026 EPS guidance is $5.23-$5.31 versus $3.77 in FY2025, a 39-41% jump.

On the fundamentals, the bear case has two real teeth: first, a US-China trade escalation that hits aerospace exports (the 10-K calls this out as a risk) would tighten demand; second, even if SpaceX never sells a third-party blade, a successful Bastrop foundry teaches Musk the process and could pressure future pricing on captive orders. The bull case, which is what Bernstein is underwriting, is that the duopoly economics — single-crystal yields that take decades to learn, customer qualification cycles that run 3-5 years, and customer agreements through 2030 — keep HWM's IGT blade business intact while the market for upstream alloys stays tight for everyone, including SpaceX.

Strip out the SpaceX headline and Howmet Aerospace's story is straightforward: a sub-segment of aerospace supply with a >50% global share, 37.7% Engine Products EBITDA margins, demand growing 38% in IGT, and contracts through 2030. The SpaceX foundry is real but aimed inward; it does not put third-party IGT blade orders at risk in this cycle. Bernstein's $328 target is not a heroic call — it's a recognition that a 9% drop on a vertical-integration story that doesn't reach the third-party market is closer to a mispricing than a regime change.