"I love constraints. In a world of constraints, you have no choice but to choose the best."
That was Jensen Huang at the Morgan Stanley Technology, Media & Telecom conference this week, explaining why ongoing shortages of AI chips and memory components are actually good news for Nvidia. He went further: "I think the fact that everything is scarce is fantastic for us."
It's a striking bit of honesty from the CEO of a company now valued at roughly $4.5 trillion. When supply is tight across the board - memory chips, advanced packaging capacity, even basic components like multilayer ceramic capacitors - customers can't afford to experiment with second-tier options. They default to whoever offers the most proven, complete solution. That's Nvidia.
Huang emphasized that Nvidia's massive balance sheet lets the company lock up supply chain capacity before competitors can. He also pitched Nvidia as the only company that can help a customer "stand up an entire AI factory" from scratch, handling everything from chip supply through deployment.
The numbers back up the confidence. Nvidia reported $130.5 billion in revenue for 2025, more than doubling an already strong 2024. The vast majority came from data center customers building AI infrastructure.
Huang also signaled he wants DRAM manufacturers to expand production capacity, essentially telling memory chip makers: build more factories, and Nvidia will buy everything they produce.
For anyone building AI applications, the subtext matters. GPU and memory scarcity means cloud computing costs for AI workloads aren't coming down anytime soon. Companies like AWS, Google Cloud, and Azure pass component costs through to customers, so Huang's enthusiasm about tight supply translates directly into higher bills for the rest of the industry. When the most powerful person in AI hardware calls shortages "fantastic," it's a reminder that the economics of running AI at scale still favor the infrastructure sellers, not the buyers.