AVGO · Q2 2025 Earnings

Management struck a more declarative tone on AI growth durability than typical semiconductor commentary. The notable shift is from prior framing — where 2025 was the visible peak with 2026/2027 dependent on customer training cadence — to an explicit statement that FY2025's ~60% AI growth rate sustains into FY2026. From the call: "we do anticipate now our fiscal 2025 growth rate of AI semiconductor revenue to sustain into fiscal 2026." This collapses the prior "post-2025 normalization" hedge that had hung over the stock.

The second shift is the rationale for that extension: inference, not training, is now driving incremental demand. Management's framing: "they are doubling down on inference in order to monetize their platforms. And reflecting this, we may actually see an acceleration of XPU demand into the back half of 2026." That pulls forward custom accelerator deployments by roughly 12 months and changes the customer mix from a handful of frontier-model trainers to a broader inference-monetization buildout. It is also why management is comfortable absorbing the Q3 FY2025 ~130bps gross margin step-down — XPU mix is dilutive to corporate margin but accretive to dollar profit at this volume.

Third, on networking: Tomahawk 6 was positioned as architectural, not incremental. "Tomahawk 6 enables clusters of more than 100,000 AI accelerators to be deployed in just two tiers instead of three." Management was clear that shipments are still proof-of-concept — production volume is an FY2026 event — but the messaging positions Broadcom as essential infrastructure rather than commodity silicon.

What management would not engage on: updated SAM math, FY2027 guidance (deferred until FY2026 is underway), and the contribution from the unnamed prospect customers. The evasion is conspicuous given the willingness to extend FY2026 commentary.