Recent reports indicate that NVIDIA has officially placed orders for Samsung’s advanced 12-Hi HBM3E memory modules. These components are expected to play a crucial role in the upcoming Blackwell Ultra rack-scale solutions, highlighting a significant collaboration between the two tech giants.
NVIDIA Partners with Samsung for HBM3E, Involves Acquisition of 50, 000 AI Chips
The long-anticipated partnership between NVIDIA and Samsung appears to be materializing. For those unfamiliar, Samsung has faced challenges in integrating into NVIDIA’s supply chain over recent months. According to a report by News1, NVIDIA CEO Jensen Huang has personally reached out to Samsung’s CEO, requesting the supply of 12-Hi HBM3E stacks tailored for NVIDIA’s Blackwell Ultra AI platforms. This development signifies that all major South Korean DRAM manufacturers are now part of NVIDIA’s supply ecosystem.
This partnership marks a significant victory not only for Samsung but also for NVIDIA, as it reinforces Samsung’s position within the highly competitive high-bandwidth memory (HBM) market. By securing this deal, Samsung is taking vital steps to reclaim its leadership in the DRAM industry.
Additionally, sources suggest that Samsung is planning to integrate 50, 000 of NVIDIA’s AI chips into the deal, although the specifics of this arrangement remain undisclosed. This points to a potential ‘circular deal’ that could unfold in the near future. Importantly, while HBM3E has established its presence in the market, attention is now shifting towards the next-generation HBM4 technology, which Samsung is reportedly poised to excel in.
There are indications that Samsung intends to offer HBM4 products at highly competitive prices, positioning itself favorably against other tech leaders like NVIDIA and AMD. However, until further details surface and the market dynamics evolve, these expectations remain speculative.
Related Articles:
AMD Launches Next-Gen RDNA GPU Architecture with Radiance Cores, Neural Arrays & Universal Compression for Enhanced Ray Tracing, Improved Upscaling, and Reduced Bandwidth Requirements
13:51
Intel’s 18A Node Hits Record Low Defect Density, Ready for Internal and External Customers
13:40
Intel Announces Transition to Annual AI Product Releases After Delays; Inference-Optimized GPU Expected to Debut by Year-End
13:39
Leave a Reply