Broadcom Gains as BofA Highlights AI Chip Cost Advantage Over Nvidia
Google's custom TPUs, designed with Broadcom, could be up to 40% cheaper to operate for certain AI workloads, according to a BofA Securities note.
Broadcom Inc. (NASDAQ: AVGO) shares climbed Monday after Bank of America Securities reiterated its Buy rating, highlighting a significant and potentially disruptive cost advantage in the high-stakes AI chip market. The firm’s analysis suggests that Google’s custom Tensor Processing Units (TPUs), co-designed with Broadcom, could be up to 40% cheaper to run for specific AI tasks than competing graphics processing units (GPUs) from market leader Nvidia.
The note underscores the growing importance of custom silicon in the AI arms race, where hyperscalers like Google are increasingly developing bespoke solutions to optimize performance and control spiraling operational costs. "This efficiency could be a major tailwind for Broadcom's custom chip business," the BofA analyst wrote, pointing to the intense demand for cost-effective AI infrastructure.
The market reacted positively to the validation of Broadcom’s strategy. While the broader market showed mixed signals, Broadcom’s stock saw increased activity, building on recent momentum. This comes on the heels of a period of strong performance, including a notable 10% stock surge in late November driven by enthusiasm for its role in Google's next-generation Gemini 3 AI model.
Broadcom, a powerhouse in semiconductor and infrastructure software, has strategically positioned itself as a key partner for tech giants looking to challenge Nvidia's dominance. While Nvidia’s GPUs have become the industry standard for training large language models, Broadcom’s expertise lies in creating application-specific integrated circuits (ASICs) that can be hyper-efficient for particular workloads. Google has been leveraging this partnership since 2016 for its TPUs, which are the backbone of many of its AI services.
Wall Street has taken notice of this accelerating AI narrative. In a separate note, Morgan Stanley also boosted its price target for Broadcom alongside Nvidia, citing the sustained momentum in artificial intelligence spending across the sector. With a market capitalization now approaching $2 trillion, Broadcom's custom silicon division is becoming a critical growth engine, projected to see revenue from AI chips grow by over 60% in fiscal 2026.
The competitive landscape is heating up as the costs associated with deploying AI at scale become a primary concern for cloud providers. As noted by Benzinga, Google's latest TPU v7 is outperforming Nvidia's Blackwell architecture in certain use cases, offering a more economical alternative that could attract more enterprise customers to Google's cloud platform.
This dynamic positions Broadcom not as a direct GPU competitor to Nvidia, but as a crucial enabler for companies building their own AI hardware ecosystems. Beyond Google, Broadcom recently expanded its collaboration with OpenAI to develop and deploy custom AI accelerators, signaling broadening demand for its specialized design capabilities.
Investors will be closely watching Broadcom's upcoming earnings report for further details on the growth trajectory of its AI segment and the financial impact of its key partnerships. The BofA report suggests that as AI workloads become more diverse, the one-size-fits-all GPU approach may be supplemented by a growing market for specialized, cost-efficient chips—a trend from which Broadcom is uniquely positioned to profit.