They have the bottleneck capacity for the 1.6T ramp and the CEO explicitly refused to raise prices to 'preserve partnership'.
Thesis: Tower is a mispriced asset: valued as a legacy analog foundry, but operating as a critical AI infrastructure bottleneck. The thesis is purely the mix shift. RF Infrastructure/SiPho jumped from 18% to 27% of revenue YoY. The 1.6T transceiver cycle is absorbing 30% of their starts. While the market worries about cyclical weakness, Tower is locking in long-term volume with Broadcom and Nvidia by trading short-term pricing power for entrenched supply chain dominance. The margin expansion is structural, driven by SiPho yields, not cycle.
Verdict: LONG — Conviction: HIGH
Catalyst: Q4 earnings confirming the $320M SiPho run-rate and first revenue shipments from the Fab 2 expansion.
Key Risk: Execution drag on the new $300M expansion pushing volume into 2H 2026, creating an air pocket in growth if legacy segments soften.
The Tell: When asked if he could raise prices, Ellwanger slipped: 'Abel is -- probably the answer to that would be yes. But considering no.' He admitted the leverage exists physically but is politically unusable against customers like Broadcom.
Friction Level: MODERATE_FRICTION — Street models a cyclical analog foundry. Management is guiding a structural shift to AI infrastructure ($900M SiPho target). The mismatch is the trade.
Report not found
The report data is no longer available. Please return to the archive.