1 Thing Broadcom Does Better Than Nvidia

Since 2022, Nvidia (NASDAQ: NVDA) has dominated the artificial intelligence accelerator market, and estimates still place its current market share at about 80%. However, other chipmakers are chipping away at its lead. Broadcom (NASDAQ: AVGO), which recently surpassed a market cap of $2 trillion, is now its biggest competitor.

Although Nvidia still offers the most popular general-purpose graphics processing units (GPUs), Broadcom's focus on custom silicon allows it to do one key thing better than its rival.

Image source: The Motley Fool.

Custom silicon creates sticky revenue

Nvidia may lead the AI chip market as a whole, but Broadcom is the leader in application-specific integrated circuits (ASICs), with about 70% market share. Broadcom serves as a design partner with hyperscalers and other AI companies that want chips custom-built to their specifications to handle highly specific types of workloads. It has deals with Alphabet, Meta Platforms, OpenAI, and Anthropic.

The development of such custom silicon is a lengthy process, requiring years of joint engineering between the AI company and Broadcom. The resulting chips are designed to fit the buyer's architecture and workload needs. Because the process is so involved, switching costs are high. If a Broadcom partner decided to switch to a different chip designer, it would need to redesign its chip.

Nvidia, on the other hand, sells standardized and versatile processors, including its current Blackwell architecture and the Vera Rubin platform, which combines both a GPU and a CPU, and is expected to ship later this year. Companies that buy Nvidia hardware aren't quite as locked into its ecosystem, though they do tend to build a lot of their software on its proprietary CUDA platform. Still, they could later switch to chips from Advanced Micro Devices, Intel, or a custom chipmaker, much more easily than Broadcom customers could move away from its custom chips.

Broadcom could lead the next phase of the AI build-out

The hyperscalers spending the most on AI infrastructure (Alphabet, Amazon, Meta, and Microsoft) are all investing heavily in custom chips. These are generally more energy efficient and cost less overall because they're built to each hyperscaler's expected needs. Given that Broadcom is the top custom chipmaker and already has partnerships with multiple hyperscalers, it stands to benefit the most from this shift in AI spending.

The company's recent results support this idea. Its AI revenue in the first quarter of 2026 was $8.4 billion, a 106% year-over-year increase. Management expects a jump to $10.7 billion in AI revenue in the second quarter, and has said that it has "line of sight to achieve AI revenue from chips, just chips, in excess of $100 billion in 2027."

The recent performances of these two semiconductor stocks also show that the market is shifting Broadcom's way. Broadcom has outperformed Nvidia over the last year, with returns of 85% through May 29, 2026.

AVGO data by YCharts

Nvidia, to be clear, has its own advantages. Its earnings are consistently excellent, and it just reported record data center revenue of $75.2 billion for its fiscal Q1 2027 (which ended April 26). That was a 92% year-over-year increase. There's still a massive market of AI companies that are sticking to off-the-shelf solutions as opposed to custom chips.

Both companies are quality investments, and I have both stocks in my own portfolio. But I'm more excited about Broadcom at the moment, and I think it will be one of the biggest winners over the next three to five years.

Lyle Daly has positions in Alphabet, Broadcom, Meta Platforms, and Nvidia. The Motley Fool has positions in and recommends Advanced Micro Devices, Alphabet, Amazon, Broadcom, Intel, Meta Platforms, Microsoft, and Nvidia. The Motley Fool has a disclosure policy.