Samsung had a tough time gaining NVIDIA's approval for HBM3 memory chips. That delay cost the company dearly, as it enabled SK Hynix to capture the majority of NVIDIA's multi-billion dollar demand.
The company has been able to turn things around for HBM4, with NVIDIA now confirming that Samsung is cleared to supply HBM4 for its next-generation AI accelerators.
All three major memory makers certified to supply HBM4
NVIDIA CEO Jensen Huang has confirmed that Samsung has been certified to supply HBM4 memory chips for NVIDIA's upcoming Vera Rubin platform.
The AI chip giant has certified all three major memory players for HBM4 supply. These include Samsung, SK Hynix, and US-based Micron. Huang confirmed that all three vendors are in production for HBM4 chips.
Samsung began mass production of HBM4 chips back in February this year. The company has also confirmed it's supplying samples of enhanced HBM4E chips to clients as well.
As the world's leading supplier of AI accelerators, NVIDIA requires a lot of memory chips. It places orders worth tens of billions of dollars with memory makers. Samsung is now well-positioned to capitalize on this demand, after missing out dearly on HBM3.
NVIDIA CEO Jensen Huang also met with the head of Samsung's foundry division earlier this week, with the latter confirming that the two companies are enhancing their collaboration on next-gen semiconductor products.
Post a Comment