Intel Unveils New GPU, Seeking Korean AI Deals
While the Nvidia ‘dynasty’ is shaking, the original processor powerhouse Intel is sending a love call to the Korean market by highlighting its design capabilities. It is expanding its artificial intelligence (AI) infrastructure product line with graphics processing units (GPUs), a market dominated b
Intel is reaching out to the Korean market with a new offering, its graphics processing unit (GPU) named "Crescent Island." The unveiling of this GPU took place at the 'AI Summit' in Seoul's COEX on November 12, replacing the spotlight on Intel's flagship central processing unit (CPU). Crescent Island is designed to facilitate high-capacity, low-power AI computing, making it a strong competitor to Nvidia's GPU products like the 'RTX Pro 6000'.
The GPU's standout feature is its memory capacity, which can reach up to 480 gigabytes (GB) using 'LPDDR5X' - a power-efficient type of general-purpose low-power DRAM. This capacity is up to five times greater than the RTX Pro 6000. Intel's Vice President of AI Products and Market Strategy, Anil Nanduri, highlighted the potential for scalability in AI inference workloads, stating that Crescent Island's expandability allows for flexibility in running models of any size.
Intel is also keen on expanding its collaboration with Korean AI infrastructure startups, offering turnkey services from chip design to production. Marcus Lawrence, Head of Custom Silicon Market Strategy at the Central Engineering Group, explained that Intel provides advanced packaging and design services, as well as foundry (contract manufacturing) services.
This presents an opportunity for Korean companies to collaborate with Intel, starting from AI chip design. The company recently formed the 'K-AI Full Stack Alliance', a cooperative system that includes 23 domestic companies, covering hardware, industry-specific services, and various other aspects of AI development.
Written by urgent.news from BusinessKorea's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
