Broadcom stock identified as a generational buying opportunity amid AI chip and networking growth

Broadcom’s shares rose sharply on April 7, gaining 6.2% while the Nasdaq Composite increased just 0.1%, as investors responded to new artificial intelligence (AI) chip and networking deals. The announcements with Anthropic and Alphabet’s Google highlighted Broadcom’s role in supplying custom AI accelerators and related infrastructure for AI data-center workloads.

Deal highlights Broadcom’s AI chip and TPU strategy

Google Cloud Services is providing multiple gigawatts of Tensor Processing Unit (TPU) capacity to Anthropic to support the scaling of its models, agents, and enterprise applications. Broadcom designs Google-built TPUs, and the company’s approach is positioned as a differentiator versus data centers that rely primarily on general-purpose GPUs.

Broadcom has argued that its custom AI accelerator solutions—referred to as XPU solutions—can ultimately outperform GPU-based designs for certain workloads. The company also has a long-standing partnership with Google, supplying application-specific integrated circuits (ASICs) engineered to handle AI workloads for Google Cloud.

Inference-focused TPU generations support AI agents

The latest TPU generations mentioned in the article—Trillium and Ironwood—are described as optimized for high-volume inference workflows, which apply what an AI model has learned. The article cites Anthropic’s Claude Code as an example of an AI agent that primarily uses inference from Claude large language models to help users develop software.

Because running AI agents requires substantial compute, the article frames the new Broadcom-linked arrangements as a validation of Broadcom’s custom chip business, particularly for hyperscalers seeking efficiency and cost advantages in AI workloads.

AI chip growth target and networking momentum

The article links the deals to Broadcom’s goal of reaching $100 billion in AI chip sales in fiscal 2027. It also notes that AI represented only a small portion of Broadcom’s business a few years ago, underscoring the scale of the expected shift.

Beyond custom AI chips, Broadcom’s AI networking business is described as expanding through its Tomahawk series (high-bandwidth switching within AI data centers) and Jericho series (AI routing to connect AI clusters across data centers). The article states that Broadcom expects AI networking to account for 40% of its upcoming second-quarter fiscal 2026 AI revenue.

How AI fits a diversified business model

While the article emphasizes Broadcom’s AI growth, it also argues that the company is not dependent solely on AI. It points to infrastructure software and general-purpose semiconductor solutions, including non-AI connectivity and storage for wireless, broadband, industrial automation, and networking.

It also highlights non-AI infrastructure software for cybersecurity and enterprise software, particularly through the VMware private and hybrid cloud platform. In the latest quarter referenced in the article, which ended Feb. 1, AI revenue made up 43.5% of total revenue.

The article suggests that AI revenue could become the majority of the business in coming quarters, but stresses that Broadcom’s non-AI segment remains a source of high-margin cash flow. That cash flow is described as supporting share buybacks and dividend increases.

Shareholder returns and valuation context

The article notes that Broadcom’s dividend yield is 0.8%, attributing the low yield to strong stock performance. It states that over the past six years, Broadcom’s dividend has doubled, while the stock is up 996%, compressing the yield.

For valuation, the article says Broadcom shares are down nearly 20% from their all-time high and trade at about 29 times forward earnings. It also cites 15 consecutive years of dividend increases and ongoing buybacks as factors supporting the investment case.