Jensen Huang says a design flaw with Nvidia's Blackwell chips has been fixed — with TSMC's help
- Nvidia's much-anticipated Blackwell AI chip appears to be back on track.
- CEO Jensen Huang said Nvidia had fixed a design flaw with the chip with the help of Taiwanese tech firm TSMC, per Reuters.
Nvidia seems to have fixed a design flaw in its next-gen AI chip, thanks to a little help from TSMC.
CEO Jensen Huang said on Wednesday the chip giant and its Taiwanese partner had fixed an issue in its much-anticipated Blackwell AI chips, per a report from Reuters.
That will come as a relief to tech giants like Meta and Google, which are hungry for Nvidia's chips to build more powerful AI systems. Morgan Stanley analysts estimated this month that Blackwell was sold out for the next 12 months.
"We had a design flaw in Blackwell. It was functional, but the design flaw caused the yield to be low. It was 100% Nvidia's fault," said Huang, in comments reported by Reuters.
The Nvidia boss said that TSMC, which is the world's largest chipmaker, helped the tech company recover from that yield deficiency and resume manufacturing of Blackwell "at an incredible pace."
Huang also reportedly denied media reports that the delays in Blackwell production had caused tension between Nvidia and TSMC, calling them "fake news."
Reports had previously suggested that the problems with Blackwell were due to a new manufacturing process employed by TSMC.
Nvidia has ridden the fever-pitch excitement over AI to become one of the world's most valuable companies, with its H100 chips in intense demand.
The company revealed Blackwell, its follow-up to the H100, in March, with Huang saying the new AI chip will be twice as fast as its predecessor.
TSMC's intervention appears to have put Blackwell's production back on track. Huang told a Goldman Sachs conference in September that the chip was in full production, and shipping was set to start in Q4.
TSMC recorded bumper earnings last week, including a 54% rise in third-quarter net profit, raising hopes that the AI chip boom still has room to run.
A spokesperson for Nvidia declined to comment beyond Huang's remarks.