NEW YORK, March 10, 2026, 09:14 EDT
Broadcom drew attention again Tuesday, following a Monday close at $345.75, up 4.6%. The chipmaker stood out among top performers on the Philadelphia semiconductor index. Gains accelerated late, continuing momentum sparked by last week’s optimistic AI forecast.
Broadcom’s ambitions are suddenly in sharper focus: last week, the company projected its AI chip revenue could break $100 billion by 2027, pulling it directly into the ring with Nvidia for data-center dominance. According to Melius Research, cited by Reuters, Broadcom already sees about 10 gigawatts worth of AI demand lined up for 2027 from customers like Meta and Anthropic. That’s comparable to the electricity use of over 8 million U.S. homes.
Alphabet, Microsoft, Amazon and Meta are on track to drop at least $630 billion into AI infrastructure this year, fueling heavier appetite for chips, servers and network hardware. Broadcom’s first-quarter AI revenue more than doubled—up 106% to $8.4 billion. Total revenue was $19.31 billion, a 29% jump. For the current quarter, Broadcom is guiding for around $22 billion, while its board signed off on a fresh $10 billion buyback, according to a March 4 filing.
CEO Hock Tan told the earnings call, “Our visibility in 2027 has dramatically improved.” Tan also pointed to “line of sight” for over $100 billion in AI chip revenue that year, crediting custom AI chip and networking demand. Reuters
Analysts wasted no time. Gil Luria at D.A. Davidson labeled Broadcom’s outlook “very encouraging.” Joseph Moore over at Morgan Stanley bumped his price target up to $470 from $462 following what he called a “strong” quarter—AI gave results a lift, networking did better than expected. Reuters
This isn’t just about a single earnings surprise. Broadcom typically doesn’t produce complete AI chips—its bread and butter is working alongside clients to turn early concepts into application-specific integrated circuits, or ASICs, making custom processors for specific workloads. Nvidia keeps its grip on the AI accelerator market, but Broadcom has become a go-to for big customers after bespoke silicon.
The playing field keeps getting wider. Just last week, Marvell projected its fiscal 2028 revenue would beat Wall Street’s targets, pointing to strong appetite for custom silicon and interconnect equipment powering AI data centers—a sign that supplier demand is spreading out. Looking ahead to next week’s Nvidia GTC conference, UBS analyst Timothy Arcuri said Monday that it’s likely to bolster views on “networking leadership, and AI capex durability.” Reuters
Broadcom’s numbers say plenty about what’s driving the story. The semiconductor solutions segment shot up 52% to $12.5 billion for the quarter. Infrastructure software? Up just 1%, landing at $6.8 billion. It’s the chip business that’s putting points on the board, not software.
Even so, it’s hardly a straightforward bet. Investors are skeptical: can this surge in AI investment really support tech’s lofty multiples? Broadcom’s fielding questions about customers rolling their own silicon, too. Tan insisted Meta’s MTIA roadmap remains “alive and well,” but cooled expectations for rapid adoption of optical links inside AI racks. For some configurations, he argued, copper still wins out on cost, power, and latency. Reuters
Broadcom’s narrative has streamlined since last month: surging AI revenue, steady networking, and management claiming supply is locked through 2028. Monday’s rally isn’t closing the argument on AI spending, but it does suggest investors are once more opting for Broadcom as a top pick outside Nvidia for custom AI chips and networking.
