Nvidia’s Hidden $60 Billion Business Is About to Overtake Broadcom

For years, investors treated artificial intelligence as a GPU story. Buy the chipmakers, ride the boom, and call it a day. But AI data centers have evolved into something much bigger — sprawling digital factories that need not only computing power, but also ultra-fast networking capable of moving oceans of data with almost no delay. 

That shift has elevated companies like Broadcom (NASDAQ:AVGO | AVGO Price Prediction), Arista Networks (NASDAQ:ANET), and Cisco Systems (NASDAQ:CSCO) into critical infrastructure suppliers. Surprisingly, though, the company best positioned to dominate AI networking may already be the same one investors still mostly think of as a GPU maker: Nvidia (NASDAQ:NVDA).

Nvidia Quietly Built an AI Networking Empire

AI data centers are bottlenecked less by raw computing power and more by how quickly thousands of GPUs can communicate with each other. That is why networking hardware has become one of the hottest battlegrounds in tech.

Broadcom’s Tomahawk switches and high-speed interconnect chips have become staples inside hyperscale AI clusters. Arista Networks has built a lucrative business connecting massive cloud data centers. Cisco, meanwhile, remains a heavyweight in enterprise networking and AI-ready infrastructure.

Here’s what the numbers look like today:

That table contains the real story. Nvidia disclosed that fiscal Q1 networking revenue reached $15 billion, tripling year over year. Annualized, that places Nvidia’s networking business at a $60 billion run rate — far surpassing Arista and already nearly equal to Cisco’s entire companywide revenue.

Regardless of how you look at it, Nvidia is no longer just a GPU stock.

The Full-Stack Strategy Is Paying Off

Nvidia’s transformation did not happen overnight. CEO Jensen Huang spent years building Nvidia into a full-stack AI infrastructure provider. GPUs were only the foundation. The company layered on CUDA software, AI development tools, networking hardware, data center systems, and now entire AI factories.

Its Mellanox acquisition in 2020 increasingly looks like one of the smartest deals in tech history. Mellanox gave Nvidia control of InfiniBand networking — technology designed to move enormous amounts of data between GPUs with minimal latency.

That matters because AI workloads break down when networking slows. A cluster of 100,000 GPUs is useless if the data traffic jams halfway through training a model.

Nvidia’s Spectrum-X Ethernet platform now directly competes against Broadcom and Cisco offerings while integrating tightly with Nvidia GPUs and software. In short, Nvidia is selling the entire AI data center instead of just one component inside it.

Granted, there are risks. Broadcom remains deeply entrenched with hyperscalers and custom AI silicon customers. Cisco still dominates enterprise networking. Arista continues growing rapidly in cloud infrastructure.

But Nvidia has one advantage none of them fully possess — control over the entire AI stack.

Why the Stock May Still Be Mispriced

Nvidia stock appeared trapped in a range for months after August despite AI spending continuing to rise. Recently, however, shares broke out to hit a new all-time high near $236 before pulling back modestly. That move may reflect investors finally understanding that Nvidia’s growth is broader than GPUs.

If networking revenue continues tripling year over year, Nvidia’s networking segment alone will soon surpass Broadcom’s entire revenue stream. That would have sounded absurd two years ago. Yet the company is already operating at a $60 billion annualized networking pace.

That also changes how investors should value Nvidia. The market often compares Nvidia strictly against semiconductor peers. But increasingly, it competes across multiple industries simultaneously — networking, cloud infrastructure, AI software, and integrated systems.

When all is said and done, Nvidia may deserve a premium valuation precisely because it is no longer a single-category business.

Key Takeaway

In short, Nvidia is evolving from the world’s leading AI chipmaker into the backbone of the AI data center itself. Its $15 billion quarterly networking revenue shows the company is capturing one of the most valuable layers of AI infrastructure at the same time competitors are fighting over individual pieces.

That said, volatility comes with the territory. Nvidia trades at elevated valuations, and any slowdown in AI spending could pressure shares in the near term.

Still, sharp investors should recognize what the numbers now make clear: Nvidia is no longer simply competing with Broadcom, Cisco, or Arista. It is increasingly becoming all three at once.