Broadcom on track for $8B+ AI revenue in 2026 driven by custom OpenAI ASIC and Ethernet switching

Analyst estimates compiled across Q1 earnings revisions now place Broadcom's 2026 AI-attributable revenue above $8 billion — roughly double the 2025 figure. Two factors dominate: the custom OpenAI inference ASIC (in design at TSMC) and the Tomahawk/Jericho Ethernet switching that lets hyperscalers wire thousands of accelerators into single training clusters.

Broadcom has become the second-largest AI-revenue beneficiary after NVIDIA, despite not making the marquee accelerator chips. The networking layer — long-running underdog versus InfiniBand — has flipped to default-Ethernet at hyperscaler scale, and Broadcom owns the merchant silicon that makes that work.

The custom ASIC project for OpenAI is the more strategically interesting line item. If OpenAI ships its own designed-by-Broadcom inference chip in 2027, it joins Google (TPU), Amazon (Trainium/Inferentia), and Microsoft (Maia) as a hyperscaler with first-party silicon. The merchant accelerator market would correspondingly contract.

Broadcom — Q1 2026 earnings call → · Fool — Cerebras IPO context →