Broadcom AI Revenue Targets Stir Investor Debate

Broadcom Inc. (Nasdaq: AVGO) reported a record fiscal third quarter in 2026, driven by a surge in AI semiconductor revenue. On Sept. 2, the company raised its AI guidance and set ambitious multiyear revenue targets. However, rising AI hardware mix and software integration pressures compressed margins, dividing investor sentiment on the stock’s outlook.

Quarterly Results and Guidance

Broadcom posted consolidated revenue of about $29.6 billion for fiscal Q3 2026, up roughly 86% year over year, with adjusted non-GAAP earnings per share of $3.32. The company guided fiscal Q4 consolidated revenue to approximately $34.8 billion, expecting AI semiconductor revenue near $21.7 billion and non-GAAP operating margins in the mid-60% range [source:8].

AI semiconductor sales reached $16.7 billion, rising 221% year over year and about 54% sequentially. These sales accounted for roughly 56% of total revenue. Shipments of XPU accelerators more than tripled year over year, representing about 73% of AI revenue. The broader semiconductor segment generated approximately $20.8 billion, up 127% year over year [source:6].

The quarter produced record free cash flow of $13.7 billion, about 46% of revenue. Operating income rose 92% year over year to a record $20.1 billion. Gross margins hovered near 75%, while non-GAAP operating margins reached the high 60% range, reflecting margin pressure from the growing AI hardware share [source:12].

AI Roadmap and Market Reaction

Broadcom raised its fiscal 2026 AI semiconductor revenue guidance to $58 billion from a prior outlook near $56 billion. For fiscal 2027, it set a target near $115 billion and, for the first time, provided a fiscal 2028 target near $230 billion. Management said supply for 2027 is secured and there is “line of sight” to 2028, with demand exceeding the current outlook [source:2].

CEO Hock Tan linked the multiyear targets to customer deployments totaling about 30 gigawatts of AI infrastructure through 2027–28. He described a revenue-per-capacity framework equating to roughly $120 billion per 30 gigawatts, illustrating how data-center buildouts support the revenue trajectory.

Broadcom’s acquisition of VMware also featured in the discussion. Management and analysts view VMware’s infrastructure software business as a source of recurring cash flow that can fund hardware investments and support a private AI cloud strategy.

Investor and analyst reactions were mixed. Supporters highlight Broadcom’s position as a fee-collecting provider of custom AI infrastructure, combining multi-generation chip programs with recurring software revenue and strong operating leverage. Skeptics point to margin compression from the heavier AI hardware mix and warn that hyperscale customers may multi-source custom ASIC projects, raising questions about the sustainability of the multiyear targets. The stock declined following the report.

Customer concentration fueled debate. Six XPU customers, including Google, Anthropic, and OpenAI, accounted for much of the AI revenue. Meta Platforms and other large cloud providers also deploy Broadcom-designed custom AI silicon, supporting the targets but concentrating exposure to a small group of hyperscalers.