Nvidia (NVDA +6.26%) delivered yet another blockbuster earnings report on May 20. The chip giant continues to rapidly grow its sales and earnings and distribute more cash to shareholders through buybacks and a 2,400% increase in its dividend. But Nvidia’s stock price has fallen since its earnings report as investor attention shifts toward the boom in artificial intelligence (AI) memory chips and application-specific integrated circuits (ASICs).

ASICs, like those designed by Broadcom (AVGO +2.94%), are cost-effective at scale for repetitive, high-volume workloads such as AI inference for chatbots, agentic AI, and search algorithm optimization.

Here’s what Nvidia and Broadcom investors should look for when Broadcom reports earnings on June 3.

1. Custom chip demand

Broadcom is at the forefront of the ASIC boom. Its most high-profile partnership is with Alphabet.

For nearly a decade, Broadcom has collaborated with Alphabet-owned Google to develop Tensor Processing Units (TPUs). Google recently released two separate chips — the 8t for training and the 8i for inference. For example, the 8t could be used for training Google’s Gemini large language models (LLMs), while 8i would be good for Google Search, YouTube recommendations, and targeted ads.

On its first-quarter 2026 earnings call, Google said it would begin selling TPUs to a select group of customers. In early April, Anthropic, the maker of the Claude family of LLMs, announced an exclusive partnership with Google and Broadcom for multiple gigabytes of compute. Claude runs on Google TPUs, Nvidia GPUs, and Amazon Web Services Trainium ASICs.

In mid-April, Broadcom extended its partnership with Meta Platforms to collaborate on AI accelerator chips and co-develop and optimize Meta’s AI infrastructure. Broadcom works with Meta to design and build out its Meta Training and Inference Accelerator (MTIA) — which is yet another version of an ASIC for hyperscale applications.

Broadcom’s AI growth is largely dependent on six customers, which include Google, Meta, OpenAI, and Anthropic. So investors should pay close attention to how Broadcom is expanding its partnerships around this core, as well as any new customer additions.

Broadcom

Today’s Change

(2.94%) $13.13

Current Price

$459.90

2. AI chip revenue guidance

During Broadcom’s March first-quarter fiscal 2026 earnings call, CEO Hock Tan made the bold prediction that the company would achieve $100 billion in AI chip revenue in fiscal 2027. For context, Broadcom earned $8.4 billion in semiconductor revenue (chips plus networking) in the first quarter of fiscal 2026 (ended Feb. 1), with guidance for $10.7 billion in the second quarter, which will be reported on June 3.

While Broadcom’s custom chips don’t compete with Nvidia GPUs in all instances, there is overlap in specialized AI workflows. On the March earnings call, Tan said that AI data centers rely too much on GPUs and expects Broadcom’s custom chips to overtake GPU designs.

Investors will want to see how Broadcom progresses toward its ambitious AI chip revenue target and how it affects Nvidia’s market share.

3. AI networking growth

For the second quarter of fiscal 2026, Broadcom expects AI networking to make up 40% of total AI revenue, which is the top end of its 30% to 40% range in any given quarter. Based on Broadcom’s $10.7 billion quarterly guidance, AI chips are expected to generate $6.4 billion in revenue while AI networking rakes in $4.3 billion.

While Broadcom’s custom chips tend to capture the spotlight, it’s really the company’s chips, paired with networking devices like Tomahawk switches and Jericho routers, that enable cost-efficient AI at scale. Tomahawk switches work with AI clusters, whereas Jericho routes across data centers. On its March earnings call, Broadcom said that its next-generation Tomahawk 7 (coming in 2027) will feature double the switching bandwidth of Tomahawk 6.

Nvidia has also been rapidly growing its networking business, which includes NVLink and NVLink Switch that compete with Broadcom’s interconnects.

An elite AI stock to build a portfolio around

Unlike some pure-play AI stocks, Broadcom doesn’t depend on a single product or end market. For example, if AI chip growth slows, it can still win in AI networking. And it generates gobs of cash flow from its non-AI semiconductor and infrastructure software business units to support consistent buybacks and a 15-year streak of dividend increases.

Add it all up, and Broadcom remains a top-tier AI chip stock for long-term investors to buy because it combines rapid, diversified AI growth with a proven legacy business model.