Broadcom CEO Reaffirms AI Revenue Goals Amid Model Slowdown Concerns
Broadcom CEO Hock Tan reassured investors that AI compute infrastructure demand remains robust, standing by the company's long-term revenue targets of $115 billion in FY2027 and $230 billion in FY2028. His comments followed market anxiety caused by Anthropic CEO Dario Amodei proposing a moderate slowdown in frontier AI model development. The reassurance highlights a broader industry shift where compute demand is increasingly driven by AI model deployment and inference rather than solely frontier training. It signals to semiconductor investors that spending on AI hardware and custom networking chips remains resilient across tech giants. Tan emphasized particular optimism around inference compute, noting that Anthropic is projected to become Broadcom's largest custom chip customer by 2027, surpassing Google. Broadcom's AI portfolio includes both custom application-specific integrated circuits (ASICs) and high-speed data center networking hardware.
## BACKGROUND
AI computing spend is divided into training (building large foundation models) and inference (running deployed models to process user requests). Companies like Broadcom design custom AI accelerator chips (ASICs) and networking infrastructure that enable hyperscalers to serve AI models more cost-effectively than using off-the-shelf GPUs alone.