Polywell is expanding our products to support AI in both Cloud Computing and Edge Computing by utilizing NVidia GPU for the Cuda AI and also to have a parallel solution with Intel technology for OpenVino AI.
We are seeing Nvidia Cuda competing with Intel OpenVino + additional new Intel AI technologies integrated into its new processors.
On the high-end side, Nvidia is now leading the GPU for AI computing with its A100. Polywell is fully supporting this platform.
Mainstream GPU for AI computing - 2U System integrating with GPU Adapters for various configurations by installing up to 4 GPU
Systems for Edge AI
NVidia Cuda Platform:
Intel OpenVino platform (just several examples - the product choice is broad):
Intel is developing its Gaudi to compete with Nvidia A100, their next generation Gaudi 3 based on PCIe 6.0 will bypass the Nvidia GPU because Intel has its PCIe bus controller inside the CPU which can block the performance I/O from Nvidia to compete with them. We will see Intel would be winning Nvidia in 3 to 5 years after their Gaudi becomes mature and more software developers porting their AI apps from Cuda into the Intel Gaudi platform. We also see Intel developing the ARC Graphics card to compete with Nvidia’s GeForce Graphics, which is hard to see how it goes, but I have confidence that Intel will do better than Nvidia after PCIe 6.0 Ultra-fast bus lanes.
Intel is making the next generation PC as an AI PC by utilizing its PCIe 5.0 bus I/O to talk to the SSD under NVME 5.0 and memory I/O with DDR5, the GPU business would also be dominated after PCIe 5.0 and 6.0 afterward.
Intel has lost the market for a decade to Nvidia, and I will see the new Intel CPU bring Intel back on track to compete, I am going after Intel to develop the products on that path to bet our future into AI because Intel’s software is much stronger than Nvidia in overall.
Besides the AI, Polywell continues to focus on the embedded segment of the business with our Mini PC and ARM Android platforms. We are also adding more resources to the Server line while aiming for AI Computing with both GPU and Network I/O SDN with 200G and 400G Network I/O modules through PCIe 5.0 and PCIe 6.0 coming along.
Looking forward to talking to you again soon.