SAN FRANCISCO, March 5, 2026, 07:03 PST
- Broadcom jumped roughly 5% out of the gate after its CEO flagged AI chip sales climbing past $100 billion by 2027.
- The company posted a 29% jump in quarterly revenue and rolled out a fresh $10 billion share buyback plan, set to run until Dec. 31, 2026.
- Investors are questioning if Big Tech can keep up its AI expansion while supply constraints persist.
Broadcom shares rose roughly 5% Thursday, hovering near $333, after CEO Hock Tan projected AI chip sales would surpass $100 billion by 2027. “We have line of sight,” Tan told analysts on the call. Bloomberg
The call hit a market jittery over the escalating price of the AI buildout. Investors are now posing a straightforward question: does all this data center spending actually deliver returns, or does it just drive up expenses?
Broadcom is leaning into custom chips—single-task processors designed to work with or even swap out expensive general-purpose AI accelerators. It’s also pushing its networking equipment, less flashy but essential hardware that connects the servers fueling the AI competition.
Reuters says Alphabet, Microsoft, Amazon, and Meta are on track to pour over $600 billion into AI infrastructure this year. Broadcom, meanwhile, has been touting a multi-year pipeline, moving past its usual quarter-to-quarter focus. CEO Tan told investors the company already locked in top-tier manufacturing capacity and high-bandwidth memory—chips crucial for shuttling data to AI processors—through 2028.
Broadcom reported late Wednesday that its first-quarter revenue hit $19.311 billion, up 29%, with AI revenue more than doubling—soaring 106% to $8.4 billion. For the current quarter, the company projects revenue around $22.0 billion. The board signed off on a new share buyback plan of up to $10 billion, running through Dec. 31, 2026, and the quarterly dividend stays put at $0.65 a share. “Consolidated revenue grew 29% year-over-year to a record $19.3 billion,” CFO Kirsten Spears said. SEC
Broadcom’s software division, which houses tools for managing corporate data centers, posted growth of just 1% for the quarter. That leaves semiconductors to pick up the slack and drive performance.
The company broke its usual pattern by offering guidance that stretches well past the coming quarter, a move that caught attention. D.A. Davidson’s Gil Luria took note, commenting, “the visibility into results more than a year out is an indication of significant growth in demand.” Reuters
Broadcom isn’t always behind every detail of AI chip design. With customers like Google, it’s involved in developing tensor processing units—TPUs—and it collaborates with OpenAI on specialized in-house processors. After those early blueprints are drawn up, Broadcom steps in to convert them into manufacturable layouts, which get sent to foundries like TSMC.
The 2027 target banks on AI spending staying strong. Should major clients pull back on capex, stall rollouts, or shift more chip design internally, Broadcom’s custom-chip run-rate might fall short of what the company’s currently projecting.
Nvidia remains the top player for AI accelerators powering big model training and inference. AMD, for its part, is moving further into data-center AI. Broadcom wants to stretch its reach—leaning on custom silicon and networking—but faces pressure to prevent its software segment from weighing down performance.
