Broadcom’s AI Ambition: CEO Projects $100 Billion-Plus in Chip Revenue by 2027
Broadcom’s stock surged nearly 5% on Thursday following remarks from CEO Hock Tan, who projected that the company’s AI chip revenue could reach “significantly in excess of $100 billion” by 2027. The forecast, made during an analyst call following better-than-expected quarterly results, underscores the accelerating demand for custom silicon and networking gear to power artificial intelligence workloads.
A Staggering Growth Trajectory
Tan’s projection far exceeds current Wall Street consensus, immediately prompting analysts to revise their models upward. JPMorgan analysts, for instance, noted that Broadcom is nearing 10 gigawatts of AI infrastructure capacity across six major customers. They estimate a revenue potential of $12 billion to $15 billion per gigawatt by 2027, leading them to “conservatively” lift their AI revenue estimate for the company to $120 billion or more.
“We see it delivering ongoing cost reductions on pace with market leader Nvidia,” wrote Goldman Sachs analysts, highlighting Broadcom’s position in AI networking and custom silicon as a key enabler for the “lowest inference cost” for its hyperscaler clients.
Addressing Profitability and Supply Chain Concerns
The bullish outlook helps allay investor worries about the profitability of Broadcom’s AI business, particularly as it ships more integrated racks of chips. In his comments, Tan asserted that the company’s yields and cost structures for AI products are now aligned with the consistent, high-margin models of its legacy semiconductor businesses.
“We have gotten our yields, we’ve gotten our cost to the point where the model we have in AI will be fairly consistent with the models we have in the rest of the semiconductor business,” Tan stated.
On the persistent shortage of high-bandwidth memory (HBM), a critical component for AI accelerators, Tan provided reassurance: the company has secured both memory supply and leading-edge wafer capacity through 2028, insulating it from near-term bottlenecks.
Navigating a Competitive Landscape
With more hyperscalers designing their own chips, some investors have feared Broadcom could be sidelined. Tan countered that the immense challenge of competing against a dominant player like Nvidia actually creates a long-term opportunity for specialized, high-performance partners like Broadcom.
“Large language model makers cannot afford to have a chip that is just good enough,” Tan explained. “You need the best chips around because you’re competing against other LLM players, and most of all, you’re also competing against Nvidia, who is by no means letting their guard down.” He believes this dynamic will benefit Broadcom “over many years to come.”
Ripple Effects Across the Supply Chain
Broadcom’s confident outlook and its strategic focus on copper connectivity for AI server interconnects had immediate, divergent effects on related stocks. Shares of Credo and Amphenol, companies whose businesses align with this copper-centric approach, rallied 10% and 4%, respectively.
Conversely, providers of newer optical interconnect technology, such as Lumentum and Coherent, saw their shares fall over 4% each, reflecting a market vote for Broadcom’s current technology path.
— CNBC’s Jordan Novet and Katie Tarasov contributed reporting
