Newshot
Account

Intel Joins Forces with 23 Korean Companies to Launch "K-AI Full-Stack Alliance," Building an AI Infrastructure Ecosystem

Source: IT之家 Published: AI 中文 AI
AI summary

Intel announced the official launch of the "K-AI Full-Stack Alliance" with 23 Korean companies at the 2026 Intel AI Summit held at COEX in Seoul on November 12, according to Korean media outlet The Elec. The alliance is a broad industry consortium centered on Intel and composed of multiple Korean AI companies, aimed at building a comprehensive AI ecosystem covering the full technology stack for Korea, with members collaborating across the entire industry chain including hardware, software, industry-specific AI solutions, and services, to jointly build a sovereign AI and open AI infrastructure ecosystem for the Korean market. The first 23 companies include Korean local server manufacturers (such as KTNF), AI chip and NPU design startups (such as FuriosaAI and DEEPX), AI large model developers (such as Upstage), and cloud computing and software service providers (such as MegazoneCloud and Hancom). In addition, Intel publicly showcased the physical D270 chip from its next-generation data center inference accelerator product line, the Intel GPU D series. Official data shows that this product can deliver up to 26,200 TFLOPS of FP4 compute at 350 watts of power consumption. Compared with

Translation(model translation, not the source's own text)

Intel Partners with 23 Korean Companies to Form the "K-AI Full-Stack Alliance", Building an AI Infrastructure Ecosystem

IT之家, November 14 — According to Korean media outlet The Elec, Intel announced at the "2026 Intel AI Summit" held at COEX in Seoul on November 12 the formal establishment of the "K-AI Full-Stack Alliance" together with 23 Korean domestic companies.

The K-AI Full-Stack Alliance is a broad industry consortium centered on Intel and composed of multiple Korean domestic AI companies. It aims to build a comprehensive AI ecosystem for Korea covering the complete technology stack. Alliance members will collaborate across the entire industry chain, including hardware, software, industry-specific AI solutions and services, with the goal of jointly building a sovereign AI and open AI infrastructure ecosystem for the Korean market.

IT之家 has learned that the 23 companies in the first batch cover Korean local server manufacturers (such as KTNF), AI chip and NPU design startups (such as FuriosaAI and DEEPX), AI large-model developers (such as Upstage), and cloud computing and software service providers (such as MegazoneCloud and Hancom).

In addition, Intel publicly showcased on site the physical chip D270 from its next-generation data center inference accelerator product line — the Intel GPU D series.

Official data shows that at 350 watts of power consumption, this product can deliver up to 26,200 TFLOPS of FP4 compute. Compared with the competing 英伟达 RTX 6000 Pro, Intel demonstrated several performance advantages, including a 30% reduction in latency when running the Gemma model, a 2.2x improvement in AI agent performance, and up to a 5.9x improvement in compute performance in financial simulations.

Related Articles

  • Intel shares fall more than 30% at the open, potentially the largest single-day drop on record
  • Intel's net profit attributable to shareholders in the third quarter of fiscal 2026 was $3.32 billion, down 18% year on year
  • The Intel SGX era is entering its final countdown, with Diamond Rapids becoming the last generation of Xeon processors to support it
  • South Korea's "Nuri" fifth launch payload includes 5 LG components and the Mobilint AI SoC
  • Ming-Chi Kuo: Intel's 18A process yield in 2026Q2 is about 80%, better than expected
  • Intel abolishes the nearly half-century-old Fellow title, renaming its top researchers Distinguished Engineers

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by IT之家
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗