Broadcom faces its biggest earnings test yet Thursday as the AI chip giant aims to deliver on CEO Hock Tan's bold $5.1 billion AI revenue target for Q3. With shares up 30% this year and analyst expectations soaring, the $1.4 trillion company must prove its custom silicon strategy can sustain momentum against intensifying competition from Nvidia and growing hyperscaler demands.
Broadcom steps into the earnings spotlight Thursday evening carrying the weight of Wall Street's highest AI expectations. The semiconductor giant, now valued at $1.4 trillion, faces its most scrutinized quarter yet as investors hunt for proof that its custom chip strategy can deliver on CEO Hock Tan's ambitious $5.1 billion AI revenue target for Q3.
The stakes couldn't be higher. Broadcom's stock has nearly doubled over the past 12 months, with a 30% surge this year alone, making it one of the market's biggest AI winners alongside Nvidia. But that meteoric rise has cranked up pressure to show the AI boom isn't just a flash in the pan.
Analysts are betting on $15.83 billion in total revenue for the quarter, representing 21% growth from $13.07 billion a year ago, according to LSEG consensus estimates. Earnings per share are pegged at $1.65, but the real fireworks will come from AI segment performance.
"All eyes will turn towards any visibility of current AI Custom Silicon engagements converting into customers with high-volume ramps in sight," Cantor Fitzgerald analysts wrote in their recent buy recommendation. The firm expects increased demand signals from hyperscale giants Google and Meta, two of Broadcom's biggest custom chip customers.
The company's secret weapon lies in its XPU accelerator chips—processors designed to run specific AI workloads more efficiently and cost-effectively than Nvidia's flagship GPUs. While Nvidia dominates general-purpose AI training, Broadcom has carved out a lucrative niche building custom silicon for tech titans who want chips tailored to their exact specifications.
In Q2, AI revenue already jumped 46% year-over-year to $4.4 billion, with 40% coming from networking infrastructure that ties thousands of AI chips together in massive data centers. Tan confidently projected that figure would climb to $5.1 billion in Q3, telling investors the growth reflects continued investment from "our hyperscale partners."
