Samsung has reportedly won a contract with NVIDIA to provide the AI GPU giant with advanced 2.5D packaging.
The news is coming from TheElec, with their sources saying Samsung's Advanced Package (AVP) team will be providing an interposer and I-Cube -- its 2.5D package -- to NVIDIA. Other companies will produce the High-Bandwidth Memory (HBM) and GPU wafers, with the 2.5D packaging housing the chip dies-CPU, GPU, I/O, HBM, and others-placed horizontally onto the interposer.
Samsung calls its 2.5D packaging technology I-Cube, while TSMC calls its 2.5D packaging CoWoS (Chip-on-Wafer-on-Substrate). NVIDIA's entire fleet of A100 and H200 series AI GPUs uses 2.5D packaging, and more importantly, the monster new 208 billion transistor Blackwell B200 AI GPU uses the same advanced packaging.
- Read more: NVIDIA's full-spec Blackwell B200 AI GPU uses 1200W of power
- Read more: NVIDIA rumored to switch to Intel for advanced packaging for next-gen AI GPUs
I wrote back late last year that Samsung was reportedly hoarding advanced 2.5D packaging equipment, and it seems that is playing out in front of our eyes now that Samsung has NVIDIA as a customer for its 2.5D packaging. According to the sources, Samsung will provide a 2.5D package with four HBM chips placed on them for NVIDIA.
Samsung winning a 2.5D packaging contract from NVIDIA likely means that TSMC CoWoS capacity is "insufficient," adds TheElec. It might not sound like much, but it's a big deal: NVIDIA wants an unstoppable amount of next-generation Blackwell B200 AI GPUs and utilizing Samsung in South Korea for 2.5D packaging points to TSMC not able to supply enough of its own CoWoS advanced packaging, forcing NVIDIA to get Samsung to jump on-board.
- Read more: NVIDIA spent $10 billion on developing its next-generation Blackwell GPU
- Read more: NVIDIA CEO: our next-gen Blackwell GPU will cost $30,000 to $40,000 each
- Read more: NVIDIA GB200 Grace Blackwell Superchip: 864GB HBM3E, 16TB/sec bandwidth
- Read more: NVIDIA Blackwell AI GPU: multi-chip die, 208B transistors, 192GB HBM3E
- Read more: NVIDIA's new Blackwell-based DGX SuperPOD: ready for trillion-parameter scale for generative AI