PALO ALTO, California, April 27, 2026, 04:07 (PDT)
Broadcom started Monday with a $2.05 trillion market cap—putting it among a handful of chipmakers at that level. The surge is less about phones or laptops these days, with custom AI chips and cloud networking now driving the story. Google’s infrastructure ambitions have only pulled Broadcom in further. Last trade: $422.76, ahead of the U.S. open.
That’s become relevant now, as Broadcom’s stock run-up has pushed it further into the core holdings of popular funds right when retail investors got access to split-adjusted shares in several Vanguard ETFs. According to a Motley Fool piece published Sunday, Broadcom landed in the top 10 holdings for four out of five Vanguard ETFs that executed splits on April 21. As for Vanguard’s High Dividend Yield ETF, it now lists Broadcom as its biggest position.
The focus isn’t really on the split itself. Splitting a stock or ETF doesn’t alter its economic value. What’s at stake is whether Broadcom deserves a spot among top “picks and shovels” names in AI — as a provider of chips, networking gear, and software powering the infrastructure for large language models. According to Vanguard, the ETF splits done on April 21 were just to make share prices easier to buy, without changing total market value. PR Newswire
Broadcom’s numbers landed with plenty for bulls to chew on. The chipmaker posted fiscal Q1 revenue of $19.31 billion—a 29% jump from last year. AI semiconductor sales more than doubled, up 106% to $8.4 billion. Chief Executive Hock Tan flagged “custom AI accelerators and AI networking” as key drivers, and Broadcom is now projecting $10.7 billion in AI chip revenue for the second quarter. Broadcom Inc.
The Google tie-up stands out. Broadcom, in an April 6 filing, disclosed a multi-year deal to co-develop and supply future versions of Google’s Tensor Processing Units—those are the custom AI chips Google leans on for model training and inference. That same document noted Anthropic is in line for roughly 3.5 gigawatts of next-gen TPU-based AI compute through Broadcom and Google beginning in 2027, provided Anthropic keeps performing commercially.
Anthropic is pitching this as a big leap in capacity. “Most significant compute commitment to date,” CFO Krishna Rao said, framing the deal. The company disclosed its annualized revenue run rate now exceeds $30 billion, and notes it’s still mixing chips from AWS Trainium, Google TPUs, and Nvidia GPUs. Anthropic
This month, Google rolled out its eighth-generation TPUs, splitting the line between TPU 8t—aimed at training hefty AI models—and TPU 8i, which is focused on inference, the process where trained models handle user queries. The company says both chips target distinct tasks inside the AI Hypercomputer setup and touts improved energy efficiency along with stronger performance-per-dollar.
Broadcom isn’t just focusing on headline-grabbing AI chips—it’s also making moves in networking. On April 22, the company announced Google Cloud is adopting AppNeta by Broadcom within its Cloud Network Insights offering, a tool designed to help users track and troubleshoot network and application issues across both cloud and on-premises environments. “Observability has never been more crucial,” said Rob Enns, Google Cloud’s VP and general manager for cloud networking. Broadcom Inc.
Competition’s fierce out there. Nvidia still dominates as the top supplier for general-purpose AI graphics chips. Broadcom and Marvell, meanwhile, are battling in the custom silicon space, building hardware tailored to cloud providers’ own demands. Earlier this month, Reuters said Google had been in discussions with Marvell about fresh AI chip projects. Customers are eager to reduce their exposure to any single vendor, said Russ Mould, investment director at AJ Bell; diversifying supply helps hedge both tech and logistics risks.
Meta just padded Broadcom’s client roster. Back in April, Reuters said Meta and Broadcom struck a deal covering multiple generations of custom AI chips through 2029 — with Meta putting up an initial pledge topping 1 gigawatt of compute. Large tech players keep looking for tighter control over their chip pipelines and spending.
The risks here aren’t trivial. Broadcom’s stock price bakes in the idea that custom AI demand will keep surging. Even the company’s own filing notes that growth tied to Anthropic hinges on Anthropic staying commercially strong. There’s also the possibility that Google and other big buyers decide to spread their orders among several chipmakers, which could blunt Broadcom’s pricing leverage—even if the AI sector keeps growing.
Right now, Broadcom isn’t getting lumped in with lesser-known chipmakers. Its shares have turned into a straight wager on custom AI silicon, cloud networking, and however Google, Meta, and other hyperscalers decide to spend on infrastructure. Vanguard’s split-adjusted ETFs just make it simpler to hold that position—though there’s no real price break under the hood.
