# 在2026年台北國際電腦展，英特爾宣布一系列聚焦資料中心、AI基礎設施、雲端推論、邊緣運算及個人電腦的更新。

*genai · news · 2026-06-02 · Firstpost*

## Key points

- 英特爾推出結合Xeon處理器與SambaNova SN50 RDU的大規模推論機架級AI基礎設施。
- 鴻海將整合英特爾與SambaNova技術，打造統一的生產就緒機架，並採用CPU優化配置。
- VectorCore Compute展示使用Xeon 6、SN40 RDU及Nvidia Blackwell GPU的完全解耦AI推論堆疊。
- 英特爾新款採用18A製程技術的Xeon 6+處理器，能在32U 100kW機架中實現高達36,864核心。
- 英特爾推出Arc G系列GPU用於手持遊戲，並擴大Core Ultra Series 3在邊緣運算與機器人領域的採用。

At Computex 2026, Intel announced a series of updates focused on data centers, AI infrastructure, cloud inference, edge computing, and PCs. The company introduced new Xeon processors, unveiled partnerships aimed at industry-specific AI deployments, and highlighted progress on its Intel 18A manufacturing technology. The announcements come at a time when AI workloads are shifting from model training toward inference and agent-based applications, creating growing demand for infrastructure designed for large-scale deployment. Rack-Scale AI Infrastructure Intel showcased new infrastructure specifically designed to support inference and agentic AI workloads at scale. The platform combines Intel Xeon processors with SambaNova’s SN50 reconfigurable dataflow units (RDUs), offering a solution for the increasing demand for high-performance AI computing in data centers. Intel added that it is collaborating with SambaNova and Foxconn to bring the technology into production-ready rack systems that integrate both companies’ technologies into a unified deployment platform. Foxconn will provide system integration expertise and is also planning a CPU-focused configuration optimized for cost-efficient inference and large-scale data-processing applications. Industry Partners Showcase Advanced AI Deployments Intel also highlighted demonstrations from VectorCore Compute, a company backed by Vista Equity Partners and Cabium Capital. The firm showcased a fully disaggregated inference environment operating from a Los Angeles data center, utilizing Intel Xeon 6 processors alongside SambaNova SN40 RDUs and Nvidia Blackwell GPUs. The demonstration highlighted how multiple processing architectures can be combined to support advanced AI workloads. Beyond this, the company announced the commercial availability of its new Xeon 6+ processor family, manufactured using Intel’s 18A process technology. The processors are designed to support high-density, scale-out computing environments and can be configured as part of Intel’s rack-scale AI infrastructure offerings. Intel says that a single liquid-cooled rack equipped with Xeon 6+ processors can deliver up to 36,864 cores within 32U of compute space while operating at approximately 100 kilowatts of rack-level power. Intel also said adoption of its Core Ultra Series 3 processors continues to expand, with the chips now featured in more than 325 consumer and commercial PC designs worldwide and selected by over 130 customers for edge AI and robotics applications. The company additionally launched its new Arc G-series graphics processors for handheld gaming devices and announced partnerships with firms including Foxconn, Siemens, Hitachi, Echo Neurotechnologies, and Greenstone Biosciences to develop AI solutions for industries such as manufacturing, healthcare, and biotechnology, as it seeks to strengthen its position across the broader AI ecosystem.

**Companies:** Intel, SambaNova, Foxconn, VectorCore Compute
**Countries:** United States, Taiwan

[Read the full story on Firstpost](https://www.firstpost.com/tech/intel-unveils-next-generation-ai-infrastructure-and-xeon-6-processors-at-computex-2026-14018077.html)

---

Canonical: https://newsio.io/zh-TW/n/316c3a4d-c9ce-4c94-b3ae-abffc2c35064/2026aixeonaiintel-18a
Summarized by Newsio from Firstpost. https://newsio.io/how-it-works