Broadcom has been the most consistently positive name in Signwl's recent analysis, and the rationale has tightened materially this week.
The picks-and-shovels position. Per the May 19 weekly synthesis: "AVGO's role as the picks-and-shovels supplier across multiple hyperscaler ASIC programmes (Google TPU, Anthropic, OpenAI, Meta) becomes more important as the substitution arc accelerates." The thesis: the custom-silicon substitution from Nvidia GPUs is no longer speculative — it is showing up in spot-market data (Trainium us-ohio +332% / 7d) — and Broadcom is the only public-market vehicle for diversified exposure to that shift. Where a single-customer ASIC bet exposes the buyer to the customer (long AVGO = long Google + Meta + Anthropic + OpenAI ASIC share collectively).
Revenue trajectory. Broadcom's own guidance — $20bn AI revenue in 2025, $100bn+ projected by end-2026 — implies a 5× lift over 18 months. Even partial realisation would re-rate the multiple. Anthropic's $21bn TPU order (Broadcom / Google, 2026–2027 delivery), Meta's extended 3-year partnership (confirmed May 15), and the long-running Google TPU manufacturing relationship together fill in the backlog evidence.
The timing question. Where the analysis is more nuanced: the Anthropic $21bn TPU order is 2026–2027 delivery, not Q2 2026. Several other ASIC programmes have similar multi-quarter ramp profiles. GPU inference (i.e. Nvidia demand) remains the marginal capacity for the next 12–18 months. So while the substitution arc is directionally locked in, the magnitude of Broadcom revenue captured in any given quarter is back-end-loaded.
The composite picture: a structurally well-positioned name in the custom-silicon transition, with backlog evidence and customer breadth that distinguish it from pure-play single-ASIC plays. The principal disagreement in our analysis is timing — not direction.