At the annual Nvidia event held two weeks ago in Silicon Valley, the tech giant sought to prove that it is much more than a chip company. In fact, Nvidia is positioning itself more as an AI infrastructure company for almost every industry - from entertainment and animation studios, through chemical and medical research to heavy industry and robotics. It's not just about graphics processors, CEO Jensen Huang insists. The company also develops chips and communication switches for AI processing data centers, an operating system for AI software, software for industrial manufacturing simulations, large language models and, most recently, a model for operating robots.
Nvidia does not manufacture all of these systems, but prefers to sell its chips through third parties, such as Dell, Lenovo and Hitachi, and ensure that by controlling most aspects of the industry's hardware and software, it will always have an advantage over its rivals, or over its customers' experience in developing these systems themselves. "You don't have to buy all of this from us, but buy at least something," Huang recently said.
Will Nvidia come to dominate the automotive world?
The ambition to gain a foothold in almost every link in its customers' value chain has taken its most extreme form in the automotive industry. Nvidia has managed to win over US auto giant General Motors (GM), which failed in efforts to independently manufacture an electric vehicle. GM it is estimated, left Qualcomm, and after closing Cruise, its autonomous car and taxi development activity in which it had invested billions of dollars. GM also laid off employees at the Tel Aviv development center that was associated with the project, and signed an agreement that gives Nvidia a partnership in almost all areas of the automaker's life.
Globes spoke with Nvidia VP and general manager of the automotive team Ali Kani, about its deep penetration into brands, the differences between it and Israeli advanced driving assistance systems (ADAS) company Mobileye Global Inc (Nasdaq: MBLY) and Nvidia's future plans.
In its agreement, Nvidia will not only provide GM with AI servers for all of its activities in the organization, it will help it train its vehicles in a virtual environment before they go on a real drive and will operate AI-based robots on the production floor in GM factories. Market executives believe that GM's move indicates pressure from investors to launch an autonomous vehicle, and playing it safe with a company that leads the AI market, more than an attempt by Nvidia to dominate all of these areas, since it has never provided all of these services together in a proven way. At the same time, with all the public relations that Nvidia has received in the past two years as the world's fastest-growing AI company, no board member will oppose the move.
If that was not enough, two weeks ago Nvidia launched Halos, a safety system for autonomous vehicles that combines AI chips with its own software and cloud services - a move that could be disastrous for Israel's Mobileye. This is a repackaging of products that the company already offers, such as AI servers for training driving models (DGX), a software simulation environment for training vehicle driving models (Omniverse), a model for operating cars and production robots (Cosmos), and an autonomous vehicle management system that allows applications to be built on top of it, with the focus on safety applications.
"Doing more than Mobileye"
In contrast to Mobileye, whose product has always been a "black box" that dictates its activities (although in recent years it has gradually begun to open it up), Nvidia allows customers to customize the product. Kani believes that this open environment is the product's great added value: "We can recommend to them and show them how to design the electric vehicle so that it is a safe vehicle," he says.
Mobileye has over 10 years of experience on the roads
Kani: "We make a product that is much more complete than Mobileye. We have a training infrastructure for a safe autonomous driving model, while Mobileye does not help with this, and it is possible to test the vehicle models in virtual simulations. We have a safe chip and operating system like Mobileye, but our customers develop their own software on them, which we help to be safe. When you use Mobileye, you are using their software."
Unlike Mobieye, which began in the field of car safety and expanded into other worlds within car computing, and was even acquired by Intel to advance this vision, Nvidia enters car manufacturers from model training and computing, an advantage that exists mainly only for Tesla in this industry.
"AI together with a safety system is a basic requirement today to operate an autonomous vehicle," says Kani. "And no one has built a car platform that knows how systems, tools and data can work together to enable automakers to have safe autonomous driving software. We've been invested in this for 20 years. When we started, they said to us: 'You're a big tech company, but do you understand vehicle safety?' At that time, we weren't in that place, but we started to study the field and understand it, and today we meet the high safety standards in Europe and the US. We have filed more than a thousand patents in the field of safety."
Kani explains that the Halos safety system is now already embedded in all of Nvidia's automotive customers at various levels of adoption. Mercedes-Benz and Jaguar Land Rover, for example, use the entire value chain offered by Nvidia: a graphics processor, a safety system, AI infrastructure, a simulation for training the vehicle and an operating system. Volvo, however, has only adopted the chipset and operating system, but is developing the autonomous vehicle software itself.
Just a week after Nvidia's launch, the Volkswagen Group, which includes brands such as Skoda, Seat, Cupra and Audi, announced the further integration of Mobileye's MQB mass modular systems in millions of vehicles of its future semi-autonomous models. After the announcement, Mobileye's share price jumped 13% but has since fallen back to its original level. Since the start of 2025, Mobileye's stock has fallen by almost 25%.
Who will be the next Tesla?
Despite its declarations, Nvidia has not yet launched an autonomous model that works fully with all the hardware and software it offers from chips to the software. The only company currently doing this is Tesla, which although it works with Nvidia chips in its data centers, is developing its own Dojo chip, which is planned to replace them. For eight years, Tesla has been diligently developing its chip, operating its own training and machine learning services for its cars, and in fact currently holds the largest amount of data on autonomous vehicle driving on the roads in the US, Europe, and China. When asked about this, Kani does not deny that his products are designed to turn automakers into more Teslas. "Not just in the car, but in the entire organization, including production," he adds.
Kani sees Nvidia's move into automotive as one that could expand into other industries, turning it into a company that touches almost every aspect of the organization. "Huang has been very focused on physical AI, which includes cars, tractors, robots - all kinds of physical products that move. Automotive is just one of those markets, it’s the first one, and it’s a trillion-dollar market potential."
Is this necessarily bad news for Mobileye? Nvidia boasts a long list of customers, but industry executives say it’s hard to know how much of a one-stop shop it is for them and what value it creates relative to its costs, as it is considered expensive compared to Qualcomm and Mobileye. Ultimately, Nvidia’s takeover of more and more areas and its plans for the production floor could bring more companies into its hands, not just those struggling to develop autonomous vehicles.
Bottom line, Nvidia increased its automotive revenue by 55% to $1.7 billion in 2024. Qualcomm reported an 87% increase from the segment ($899 million), while Mobileye reported a 20% decline in revenue to $1.65 billion.
Mobileye declined to comment.
Full Disclosure: The author was a guest of Nvidia at the GTC Conference in San Jose.
Published by Globes, Israel business news - en.globes.co.il - on April 1, 2025.
© Copyright of Globes Publisher Itonut (1983) Ltd., 2025.