Prime Highlight:
- Broadcom shares surged after Google announced a sharp increase in capital spending, signaling stronger demand for AI-focused chips and data center hardware.
- Google’s emphasis on its in-house AI chips positions Broadcom as a key beneficiary due to its role in designing and building Google’s tensor processing units.
Key Facts:
- Google plans to spend up to $185 billion on capital expenditures this year, nearly double its spending from the prior year, to expand AI data center infrastructure.
- Broadcom works with Google on custom TPUs and is developing AI-specific chips for five customers, while Nvidia shares also rose on expectations of broader AI-driven demand.
Background
Broadcom shares climbed 6% in extended trading on Wednesday after Google reported earnings and announced a sharp increase in capital spending aimed at expanding its artificial intelligence infrastructure. Investors reacted positively to the scale of Google’s investment, which is expected to lift demand for advanced chips and data center hardware.
Google said it plans to spend as much as $185 billion on capital expenditures this year, nearly double its spending from last year. The company is ramping up investment to build and expand data centers that support artificial intelligence workloads, joining other major technology firms in accelerating AI-focused spending.
Technology analysts said the move is a major positive for companies tied to Alphabet’s supply chain. Ben Reitzes, head of technology research at Melius Research, described the spending plan as highly supportive for chipmakers linked to Google’s AI efforts. He noted that the scale of investment signals long-term growth for partners involved in building AI hardware.
Unlike many AI systems that rely heavily on Nvidia chips, much of Google’s AI software runs on its own tensor processing units, or TPUs. Google’s latest Gemini 3 AI model was trained using these custom chips. Broadcom works closely with Google to design and build TPUs, so it benefits directly from the higher spending.
Broadcom has been expanding its custom chip business and focusing on application-specific integrated circuits, or ASICs. These chips are designed to handle certain AI tasks more efficiently. In December, Broadcom said it would sell Google’s TPU Ironwood rack systems to AI startup Anthropic, highlighting growing demand beyond Google.
Industry experts say custom AI chips are mainly used by large cloud companies, often called hyperscalers. Broadcom said it is currently developing custom chips, which it calls “XPUs,” for five customers. Other tech giants, including Microsoft, Amazon, and Meta, are also working on their own AI chips.
Google continues to use Nvidia chips alongside its TPUs. Nvidia shares rose about 2% in extended trading, reflecting expectations that rising AI spending will support multiple players across the chip sector.
