Samsung Expects HBM3E Integration In NVIDIA’s AI Accelerators By Next Quarter, Refuting Rumors of “HBM Business” Fall-Out

Samsung’s ambitions of supplying HBM products to NVIDIA aren’t over yet, as the firm announced that their 5th-gen HBM3E memory is slated to be used in NVIDIA’s flagship AI accelerators.

Samsung’s Hopes For HBM Adoption Are Pretty Much Alive, As The Korean Giant Reveals HBM3E Supply To NVIDIA In Q4

The Korean giant was hit with a massive “business shock” when it was revealed that the firm has been unable to secure NVIDIA as an HBM customer, citing that the process has been delayed indefinitely, which created the assumption that Samsung has failed to establish its presence in NVIDIA’s supply chain. However, in its recent earnings call, Samsung revealed that its journey toward becoming an NVIDIA supplier is ongoing and that the firm is on the right track to supply 8-Hi HBM3E memory to a “major customer.”

The remarks by Samsung’s VP of Memory Business Division, Kim Jae-jun, don’t explicitly mention NVIDIA. Instead, he revealed that Samsung is in pursuit of establishing an HBM3E supply with a significant customer, and analysts say that this is likely pointing toward NVIDIA, as the Korean giant is now cautious with its statements. Here is what Samsung’s VP had to say:

Both HBM3E 8-stack and 12-stack products are being mass-produced and sold, and significant stages of the quality test process for a major customer have been completed, leading to sales expansion in the fourth quarter.

We are preparing improved HBM3E products in line with the next-generation graphics processing unit (GPU) projects of our major customer.

– via BusinessKorea

In a previous…