Even Nvidia isn't immune to the AI chip shortage.
The company's automotive division still has to compete internally for access to the GPUs that have made Nvidia the world's most valuable company, according to Xinzhou Wu, Nvidia's head of automotive.
"Even at Nvidia, basically we do have a limited supply of GPU for compute," Wu said in an episode of The Verge's "Decoder" podcast that aired on Monday.
As demand for Nvidia's chips continues to surge from AI companies building massive data centers, Wu said different teams across the company regularly compete for computing resources needed to train and test their own AI models.
"We have an internal priority, and I'm working with my colleagues basically almost on a weekly basis to decide how to set aside this different compute, sometimes for training, sometimes for test resources for different threads of work in the company," Wu said.
"And sometimes we need Jensen to help," Wu said of the company's CEO, Jensen Huang.
The comments offer a rare glimpse into how Nvidia allocates resources inside a company whose GPUs have become the backbone of the generative AI boom. Demand for its chips has consistently outpaced supply as companies, including OpenAI, Microsoft, Meta, xAI, and Amazon, race to build ever-larger AI models.
Wu said decisions aren't driven solely by near-term revenue.
"It's all of the above," he said when asked how those trade-offs are made. Nvidia balances current business needs with long-term strategic opportunities, including what Huang calls "the zero trillion dollar business" — entirely new markets that could eventually be worth trillions of dollars, Wu said.
One of those bets is autonomous driving.
Wu said Nvidia believes "everything that moves will be autonomous" and is investing heavily in supplying chips, software, AI models, simulation tools, and safety systems for self-driving vehicles. While the automotive business remains much smaller than Nvidia's booming data-center division, Huang continues to prioritize it.
"We are strong believers — Jensen himself as well — of the AV [autonomous vehicle] future," Wu said. "We are keeping investing basically in this technology and in this future, not only from allocating external compute but from fab capacity as well."
Wu also said that even semiconductor manufacturing capacity has become another internal battleground as demand for Nvidia's chips continues to soar.