Revenue grew 36% and FCF tripled, yet the street is worried about a 2026 power cord.
Thesis: The market views AMD as a 'nice to have' second source; the OpenAI 6GW partnership confirms they are a 'must have' structural necessity. Hyperscalers cannot let NVIDIA dictate unit economics forever. The ZT Systems sale (keeping design, dumping low-margin mfg to Sanmina) proves Su is protecting the margin profile while scaling the ecosystem. This isn't a pivot; it's an entrenched duopoly forming in real-time.
Verdict: LONG — Conviction: HIGH
Catalyst: Analyst Day on Nov 11 to detail the 'tens of billions' 2027 roadmap, specifically the MI450/Helios unit economics.
Key Risk: Execution risk on the 'first gigawatt' deployment in 2H 2026. If Helios misses power/cooling targets, the backlog evaporates.
The Tell: Su's answer on MI308. She admitted receiving licenses—a bullish signal—but immediately killed the revenue contribution in the guide. 'We'll have to see how the demand environment shapes up.' She refuses to let the China wild card pollute the clean execution narrative.
Friction Level: MODERATE_FRICTION — The Street sees the OpenAI deal as 'paper backlog' dependent on non-existent power. We see it as the 'Pricing Power Signal' that validates the hardware stack.
Report not found
The report data is no longer available. Please return to the archive.