Nvidia design flaw with Blackwell AI chips now fixed, says CEO

By Jacob Gronholt-Pedersen and Supantha Mukherjee

COPENHAGEN/STOCKHOLM (Reuters) – Nvidia Chief Executive Jensen Huang said on Wednesday that a design flaw in its latest Blackwell AI chips that affected production had been fixed with help from its Taiwanese manufacturing partner long-time TSMC.

Nvidia unveiled the Blackwell chips in March and had previously said they would launch in the second quarter, but they were delayed, potentially affecting customers such as Meta Platforms, Alphabet, Google and Microsoft.

“We had a design flaw at Blackwell,” Huang said. “It was functional, but the design flaw meant that the performance was low. It was 100% Nvidia’s fault.”

According to media reports, the production delay caused tensions between Nvidia and TSMC, but Huang dismissed this as “fake news”.

“To make a Blackwell computer work, seven different types of chips were designed from scratch and had to be put into production at the same time,” he said.

“What TSMC did was help us recover from this yield hardship and get Blackwell manufacturing back to an incredible place.”

Nvidia’s Blackwell chips take two squares of silicon the size of the company’s previous offering and unite them into a single component that’s 30 times faster at tasks like providing chatbots with responses.

At a recent Goldman Sachs conference, the CEO said the chips will ship in the fourth quarter.

Huang was in Denmark on Wednesday to launch a new supercomputer called Gefion, which has 1,528 graphics processing units (GPUs) and was built in partnership with the Novo Nordisk Foundation, the Danish Export and Investment Fund and Nvidia.

(Reporting by Jacob Gronholt-Pedersen in Copenhagen and Supantha Mukherjee in Stockholm; Editing by Elaine Hardcastle)