Opinion: The biggest risks facing Nvidia stock and how the company will solve them

There has always been cyclicality in the semiconductor industry, as consumers and businesses only upgrade their physical hardware once every few years, whether it’s computers, smartphones, or even data centers. That means there are big waves in earnings, followed by long troughs.

Recently, artificial intelligence (AI) has changed this. From 2023 onwards, spending on data center chips and components looks set to skyrocket as some of the world’s biggest technology companies compete to develop the most powerful AI software.

Nvidia (NVDA 2.27%) The company has been the biggest beneficiary of that spending boom because it supplies cutting-edge data center graphics processors (GPUs) for AI development. Nvidia sells so many chips that it has become the second-largest company in the entire world, increasing its market capitalization by $3 trillion in the past two years alone.

But data center spending won’t continue at this pace forever, and an inevitable slowdown could be the biggest risk to NVIDIA stock. But CEO Jensen Huang may already have a solution.

Image source: Nvidia.

Data center spending could slow significantly within a few years

The ultimate goal of all companies developing AI is to achieve artificial general intelligence (AGI). This is the point at which technology rivals human intelligence in most cognitive tasks. Researchers who worked for ChatGPT creator OpenAI predict that AGI could arrive as early as 2027. Tesla CEO Elon Musk believes 2029 is a more realistic goal. In any case, it may be just a few years away.

Developing AI beyond the level of AGI will almost certainly lead to diminishing returns. This is because few commercial workloads benefit from such advanced machine intelligence. If that’s the case, demand for Nvidia’s data center GPUs could plummet years from now, as the pool of developers who want (or can afford) further performance increases will be very small.

Today, the vast majority of spending on data center infrastructure comes from multi-trillion dollar tech giants (more on that in a moment), and Nvidia has a We are launching a new generation of data center GPUs. The company released the Hopper architecture in September 2022, the Blackwell architecture in March 2024, and the report says a new architecture called Rubin is expected to be rolled out by the end of 2025.

This is not sustainable in the long term. Not only is demand likely to slow within a few years, but NVIDIA has spent $10 billion developing Blackwell alone, and R&D costs will only rise from there.

Blackwell is in such high demand right now that Nvidia can’t keep up, so there’s nothing to worry about in the short term.

Nvidia’s growth relies heavily on just a handful of customers

Nvidia is on track to generate record total revenue of $128.6 billion during its 2025 fiscal year, which ends at the end of this month. This is a whopping 112% increase over the previous year. In the first three quarters alone, NVIDIA’s data center division accounted for approximately 88% of its total revenue, primarily from GPU sales.

As mentioned earlier, most of the infrastructure spending for AI data centers is being carried by a small number of large technology companies. That’s why in Nvidia’s most recent third quarter (ending October 27), 36% of its $35 billion in total revenue came from three anonymous customers.

Data source: Nvidia.

Morgan Stanley estimates that only four companies will spend a combined $300 billion building AI data centers in 2025: Microsoft, Amazon, Alphabet, and Meta Platforms. These names could be some of Nvidia’s mystery customers in the table above.

In my opinion, this increases the risk NVIDIA faces if AI spending slows. That’s because if one of its top three customers were to suddenly withdraw, it would leave a huge hole in the company’s revenue base that would be virtually impossible to fill. That could threaten Nvidia’s ability to maintain its impressive growth rate and lead to a significant stock correction.

Image source: Nvidia.

Solution: Nvidia significantly diversifies its revenue

Speaking at the CES 2025 Technology Conference on January 7, Jensen Huang highlighted several new opportunities for Nvidia as the AI ​​industry expands beyond the data center. Self-driving cars are one of them, and Huang says this could be the first multitrillion-dollar sector of the emerging AI-powered robotics industry.

Nvidia has been in the automotive business for over 20 years. It is now home to the Drive platform, an end-to-end solution for automakers looking to add self-driving capabilities to new vehicles. It includes all the hardware and software the manufacturer needs, including Nvidia’s powerful Thor chip, which processes data from the car’s sensors to facilitate real-time decision-making on the road.

Automotive giants like Toyota, Mercedes-Benz, Hyundai, BYD, and Volvo have already partnered with Nvidia. Apart from using the Drive platform, some of these companies are also purchasing DGX data center systems with Blackwell-based GP200 GPUs to continuously train self-driving models, Huang said. It has enough computing power to do so.

Additionally, Nvidia just announced a new multimodal foundation model called Cosmos. It has been trained on 20 million hours of video, giving it the ability to understand the real physical world. Automakers can use it to run millions of simulations using synthetic data to serve as training material for self-driving software. This is another way Nvidia is attracting customers to its ecosystem.

Nvidia plans to generate approximately $1.5 billion in revenue from its automotive division in fiscal 2025. But that number could more than triple to $5 billion in 2026, Huang said. According to Wall Street consensus forecasts (provided by Yahoo!), Nvidia will achieve $196 billion in total revenue in fiscal year 2026, so the auto division will still be very small.

However, if it grows consistently at this pace over the next few years, it could become a key part of Nvidia’s business in time for a potential slowdown in data center spending.

Furthermore, Cathie Wood’s Ark Investment Management believes that self-driving ride-hailing services, just one use case for self-driving cars, will generate $14 trillion in enterprise value by calendar year 2027. The company believes most of its value will come from its self-driving platform. With providers like Nvidia also on board, the opportunity could be even bigger than Huang currently anticipates.