Broadcom (NASDAQ: AVGO) has emerged as the dominant supplier of nearly everything inside a hyperscale AI cluster except the GPU, making it a compelling investment target.
Every time analysts examine how Google, Meta, OpenAI, and Anthropic build compute infrastructure, Broadcom’s name appears on custom accelerators, Ethernet switches, DSPs, optics, and fabric components.
The core investment thesis is straightforward: Broadcom captures the hyperscaler AI capital budget that does not flow to Nvidia (NASDAQ: NVDA), owning both the custom silicon and AI networking interconnect markets.
CEO Hock Tan’s team has built a formidable position across XPUs and ASICs, representing two of the three critical legs of the modern data center alongside GPUs.
The financial receipts behind this conviction are striking, with AI semiconductor revenue jumping from $5.20 billion in Q3 FY25 to $10.80 billion in Q2 FY26, a 143% year-over-year increase.
Q3 FY26 guidance calls for $16.0 billion in AI semiconductor revenue, representing growth of over 200% year-over-year, with management targeting $56 billion for full-year FY26 and in excess of $100 billion by FY27.
Hock Tan stated that “Demand for XPUs and networking is simply insatiable,” a claim supported by the $30 billion-plus in Q2 AI bookings recorded during the quarter.
The company’s margin structure is exceptional, with Q2 FY26 free cash flow reaching $10.26 billion, representing 46% of revenue and a 60.07% year-over-year increase, while adjusted EBITDA landed at 69% of revenue.
Net income grew 87.51% to $9.31 billion, funding both a $10 billion buyback authorized through December 2026 and a dividend raised 10% last December to $0.65 per quarter, marking the 15th consecutive annual increase.
Networking made up almost 40% of AI revenue in Q2, with Broadcom’s Tomahawk 6 standing as the industry’s only 100-terabit Ethernet switch, and a 200-terabit successor taping out this quarter.
The customer commitment pipeline is extraordinary in scale, with Anthropic committing to over 1 gigawatt of TPU compute in 2026 and another 5 gigawatts starting 2027, while OpenAI is contracted for 1.3 gigawatts in 2027 inside a 10-gigawatt by 2029 deal.
Meta signed for 3 gigawatts of MTIA XPUs through 2028, representing multi-year, multi-generational commitments that span several product generations and anchor Broadcom’s revenue visibility.
The exposure Broadcom provides that Nvidia does not is direct access to customers actively building alternatives to Nvidia, including Google’s TPU roadmap, Meta’s MTIA, OpenAI’s custom silicon, and Anthropic’s compute infrastructure.
Customer concentration remains a genuine risk, as a handful of hyperscalers drive the AI revenue number, though the Q2 call disclosed $6 billion in orders from two additional core customers shipping in late 2026, broadening the base.
At $418.16, the stock trades at roughly 65x trailing and approximately 21x forward earnings, with analyst price targets averaging $527.88, reflecting the market’s recognition of its exceptional compounding characteristics.
A company generating 69% EBITDA margins on an AI business projected to grow from $56 billion this year toward $100 billion next year represents a rare category of compounder, one whose customers plan their power grids around its ship dates.
