Nvidia is expanding beyond its core GPU business into custom silicon, creating a new $60 billion growth engine that directly competes with Broadcom's primary strategy.
Back
Nvidia’s $60B Custom Chip Unit to Challenge Broadcom’s Lead
Nvidia Corp. is leveraging its massive data center success to build a custom chip design business reportedly valued at $60 billion, a move that pits it directly against Broadcom Inc. in the lucrative market for bespoke AI accelerators. The initiative targets the same hyperscale customers that Broadcom serves, creating a new competitive front in the battle to supply the world’s AI infrastructure. This expansion comes as Nvidia’s data center revenue soared 92 percent year-over-year to $75.2 billion in the last quarter.
Broadcom’s management has expressed confidence that its custom AI chip segment could generate $100 billion in revenue in 2027 by co-designing processors with clients. "Instead of designing broad-purpose GPUs like Nvidia, it's partnering directly with hyperscalers to design AI chips that are customized to handle the specific workloads the buyers expect them to see," noted Keithen Drury, a contributor at The Motley Fool. This highlights the strategic divergence that is now blurring as Nvidia enters the custom design space.
Nvidia’s push is built on the foundation of its near-total dominance in the market for general-purpose AI GPUs, a business that accounted for the vast majority of its record-breaking quarterly revenue. In contrast, Broadcom has carved out a niche by creating application-specific integrated circuits (ASICs) for tech giants who want to optimize performance and cost for their unique AI models. Advanced Micro Devices Inc. is also a formidable player, with its data center revenue jumping 57 percent to $5.8 billion in its most recent quarter.
The strategic clash is for a piece of a rapidly expanding pie, with Nvidia’s leadership projecting the global data center capital expenditure market to grow from $600 billion in 2025 to as much as $4 trillion by 2030. For Nvidia, success in custom silicon would solidify its reign over the entire AI hardware stack. For Broadcom and AMD, winning major custom deals is critical to proving that viable, high-performance alternatives to Nvidia's GPUs can capture a significant share of this generational investment opportunity.
A Reporting Shift Masks Growth
Nvidia recently altered its financial reporting structure, consolidating its gaming, professional visualization, and automotive segments into a single "Edge Computing" division. This new unit reported $6.4 billion in revenue last quarter. While the company stated the change "better reflects its current and future growth drivers," it obscures the performance of individual business lines.
The previous structure provided a clearer, if still combined, view of its gaming and AI software business. The new framework makes it more difficult for investors to parse the growth drivers within the 10 percent of the business that isn't data centers. Regardless of the reporting change, the sheer scale of the data center division, which grew 21 percent from the previous quarter, provides the financial and technical firepower for Nvidia to attack new markets like custom silicon.
The Hyperscaler Battleground
The primary customers for these powerful chips—hyperscalers like Meta Platforms, Microsoft, and Google—are actively diversifying their silicon supply chains to reduce reliance on any single vendor and to optimize for specific workloads. This trend is a significant tailwind for both Broadcom and AMD.
AMD, for its part, has secured major commitments that underscore this demand. In February, the company announced a multi-year deal with Meta for its Instinct GPUs, following a similar large-scale agreement with OpenAI disclosed last October. AMD CEO Lisa Su now has "strong and increasing confidence" the company can achieve tens of billions of dollars in data center AI revenue in 2027, signaling fierce competition ahead.
For investors, Nvidia's entry into the custom chip market represents a major new growth vector beyond its already dominant GPU business. While its stock trades at a premium, this expansion could help justify its valuation, which sits at a forward price-to-earnings ratio in the forties. Meanwhile, AMD's stock trades at a forward P/E multiple of around 59, reflecting high expectations for its ability to capture market share. Broadcom's strategy has been rewarded by the market, and the direct challenge from Nvidia will be a critical test of its leadership in the custom ASIC space.
This article is for informational purposes only and does not constitute investment advice.
[1] Nvidia's Hidden $60 Billion Business Is About to Overtake Broadcom[2] Nvidia is so dominated by AI, it no longer sees gaming GPUs as a standalone section of its business[3] A Once-in-a-Generation Opportunity: 5 Artificial Intelligence (AI) Stocks Primed for Massive Upside[4] Meta's $125 billion AI bet might be the capex trap of the decade[5] After a Q1 Blowout and a Soaring Share Price, Is AMD Stock Still a Buy?
