NVIDIA’s Jensen Huang Admits Blackwell AI Chip Design Error, TSMC’s Aid

CEO of NVIDIA Jensen Huang admits that design flaws in Blackwell AI chips were “100% NVIDIA’s fault” and talks about how important TSMC’s help was during the launch.

Jensen Huang, CEO of NVIDIA, talked about the issues with the Blackwell AI, the company’s newest chip.He said that there was a design mistake that was “100% NVIDIA’s fault” and that TSMC helped them get through the tough Blackwell AI GPU launch.

NVIDIA first showed off its new Blackwell chips at GTC 2024 in March of this year. They were supposed to ship in Q2 2024, but they were late (as you can see in the stories below), which could have hurt Meta, Microsoft, and Google, which all pay a lot.

Huang told us that there was a mistake in how Blackwell was made. It worked, but the output was low because of a mistake in the design. It was entirely NVIDIA’s fault. Seven different kinds of chips had to be made from scratch and rushed into production at the same time so that a Blackwell computer could work.

He said: “What TSMC did, was to help us recover from that yield difficulty and resume the manufacturing of Blackwell at an incredible pace” .

Jensen’s words come right after rumours that NVIDIA’s problems with its Blackwell AI chips hurt its relationship with TSMC and that Samsung could make the GeForce RTX 50 series “Blackwell” gaming GPUs. Of course, that is crazy, and now the CEO of NVIDIA has come out and cleared the air… and TSMC of any blame for Blackwell’s design flaws.