Nvidia (NVDA -1.50%) was founded in 1993, and it went on to create the world's first graphics processors (GPUs) for computing, media, and gaming applications. Now, decades later, the company has adapted those powerful chips for data centers, where they are used to develop advanced artificial intelligence (AI) models.
Nvidia CEO Jensen Huang believes data center operators will spend $1 trillion over the next four years on upgrading their infrastructure to meet demand from AI developers. Since the data center segment currently accounts for 88% of Nvidia's total revenue, that spending will be instrumental to the company's future success.
However, the semiconductor industry has always been cyclical, so the data center boom won't last forever. That's why it's critical for Nvidia to diversify its revenue streams, and at the CES 2025 technology conference on Jan. 7, Huang delivered some incredible news for investors on that front.
Image source: Nvidia.
Meet Nvidia's next multitrillion-dollar opportunity: Autonomous vehicles
Nvidia saw the autonomous driving revolution coming. In fact, the company's automotive business is more than two decades old, but its revenues were so tiny that it lived in the shadow of the gaming and data center segments. That's all about to change, because global car brands like Mercedes-Benz, Hyundai, BYD, Volvo, Toyota, and more are adopting Nvidia's Drive platform to power their autonomous ambitions.
Drive provides all of the internal hardware and software a car needs for self-driving capabilities. That includes Nvidia's latest chip called Thor, which processes all of the incoming data from the car's sensors to determine the best course of action on the road. But Nvidia's opportunity doesn't end there, because it also sells the infrastructure a car company needs to maintain and improve its autonomous models, so it can differentiate itself from the competition.
In addition to Drive, Huang says car companies are buying DGX data center systems featuring its latest Blackwell-based GB200 GPUs, which deliver the necessary computing power to continuously train self-driving software. Then there is Nvidia's new Cosmos multimodal foundation model, which allows companies to run millions of real-world simulations using synthetic data, serving as training material for the software.
Overall, Huang says autonomous vehicles could be the first multitrillion-dollar opportunity in the emerging robotics space. He's not alone, because Cathie Wood's Ark Investment Management thinks technologies like autonomous ride-hailing could create $14 trillion in enterprise value by 2027, with the majority of that value attributed to autonomous platform providers -- in this case, that would be Nvidia.
Nvidia's fiscal year 2025 will finish at the end of January, but the company generated $1.1 billion in automotive revenue through the first three quarters (if we extrapolate that result, full-year revenue will probably be around $1.5 billion). Huang says in fiscal 2026, Nvidia's automotive revenue could soar to $5 billion, so it's going to ramp up insanely fast.
Data center chips are the main story for now
Wall Street's consensus forecast (provided by Yahoo) suggests Nvidia could generate a whopping $196 billion in total revenue during fiscal 2026, so the automotive segment's potential $5 billion contribution would still be relatively tiny. It's a longer-term story which could secure Nvidia's future growth, but in the here-and-now, it's all about the data center.
Nvidia just started shipping its new Blackwell GB200 GPUs to customers, but sales are expected to grow quickly. By April this year, revenue from Blackwell chips could overtake revenue from the previous generation of chips built on the Hopper architecture, which highlights how quickly Nvidia's business is evolving.
The GB200 NVL72 system is capable of performing AI inference up to 30 times faster than the equivalent H100 GPU system, so Blackwell will pave the way for the most advanced AI models to date. Therefore, over the next year or so, consumers and businesses might have access to the "smartest" AI software applications (like chatbots and virtual assistants) so far.
Demand for Blackwell chips is outstripping supply, which should support further strength in Nvidia's revenue and earnings during fiscal 2026. Plus, some reports suggest a Blackwell successor called "Rubin" might be unveiled later in the year, which would further cement the company's chokehold on the market for data center GPUs.
It's probably not too late to buy Nvidia stock, despite its incredible run
Nvidia stock has soared by 830% since the start of calendar year 2023, lifting the company's value from $360 billion to an eye-popping $3.3 trillion in just two years. Despite the amazing run, the stock might still be cheap.
It currently trades at a price-to-earnings (P/E) ratio of 53.6, which is a discount to its 10-year average P/E ratio of 59. But Wall Street's consensus estimate suggests Nvidia could generate $4.44 in earnings per share in fiscal 2026, placing its forward P/E ratio at just 30.6.
In other words, Nvidia stock would have to soar by 92% over the next 12 months just to trade in line with its 10-year average P/E ratio of 59.
Nvidia has a habit of beating Wall Street's forecasts, so it's possible the stock has even more upside potential. On the flipside, there is some competition emerging from other chip makers like Advanced Micro Devices, which plans to release a Blackwell rival in a few months. That's a risk investors should keep an eye on as this year progresses.
