# 英特爾在 Computex 貿易展發表新一系列資料中心技術。

*genai · news · 2026-06-01 · Times of India*

## Key points

- 英特爾全新 Xeon 6+ 處理器採用先進的 Intel 18A 製程技術，專為資料中心打造。
- Xeon 6+ 晶片提供即時應用程式能量遙測，讓企業能監控每個程式的電力使用情況。
- 英特爾宣稱一台 Xeon 6+ 伺服器可取代多達九台舊伺服器，大幅降低能源與空間需求。
- 全新 Intel Ethernet E835 網路卡提供 200Gbps 傳輸速度，能效比 Nvidia ConnectX-6 DX 高出最多 1.9 倍。
- 英特爾即將推出的 Crescent Island GPU 專為記憶體與電力密集型的代理式 AI 系統優化，不僅限於 AI 訓練。

Intel has announced its new range of data center technologies, including processors, designed for the next generation of AI software as the company aims to claim a prominent space in the artificial intelligence universe. The announcements, made at the Computex trade show, also include high-speed networking gear and a sneak peek at a new graphics chip code-named "Crescent Island." These chips are capable of handling autonomous “agents” that can execute multi-step tasks on their own even as the traditional CPU is re-emerging as the brain that coordinates it all. Intel Xeon 6+ processors features Intel’s headline announcement is the Xeon 6+ processor family, which focuses heavily on cutting down the massive energy bills that plaque modern AI data centres. It marks the first time Intel is using its cutting-edge “Intel 18A” manufacturing process for a data centre chip. The new chips are designed to keep applications responsive under real-world power constraints, the company claimed. Their key features include up to 288 “Efficient-cores”, delivering up to 2.5 times better performance than older models, while offering 45% better power efficiency per thread than rival chips. Intel also claims the new chips are so powerful that businesses can replace up to nine older servers with just one new Xeon 6+ server, shrinking a data center's physical footprint and lowering energy costs. Thirdly, a new built-in feature called Application Energy Telemetry allows companies to monitor exactly how much electricity individual software programs are consuming in real time. The new processors are already being built into systems by tech giants like Dell Technologies, Lenovo, HPE, Asus, and Supermicro. Intel also broadened the family for small and mid-sized businesses by introducing a smaller, budget-friendly 12-core model. Breaking network bottlenecks with the E835 As AI systems scale up, moving data between thousands of computer chips quickly has become one of the tech industry’s biggest hurdles. To tackle this, Intel expanded its 800 Series Ethernet portfolio with the new Intel Ethernet E835 controllers and network adapters. These devices can supercharge data speeds up to an ultra-fast 200 gigabits per second (200GbE). According to Intel, the E835 network adapter is up to 1.9 times more energy-efficient than Nvidia’s comparable ConnectX-6 DX card and 1.4 times more efficient than a similar offering from Broadcom. Next-gen AI: The 'Crescent Island' teaser Looking to the future, Intel shared early technical details about its next-generation data center GPU, code-named Crescent Island. While standard graphics chips were built for heavy AI training, Crescent Island is being specifically optimized to run “agentic systems.” These are autonomous AI agents that require intense memory and quick thinking rather than just raw math power. Intel says the upcoming chip will directly solve the memory and power bottlenecks that currently cause AI tools to stall or lag. End of Article Latest Mobiles View All Itel A100 Pro 4G OPPO A6c 4G HMD Vibe2 5G ₹10,999 Realme 16T ₹29,999 Motorola Razr Fold ₹1,49,999 Ai+ Nova Flip ₹39,999 OnePlus Nord CE 6 Lite 5G ₹31,999 Moto G37 Power 5G Moto G37 5G ₹17,999 Realme 50A Prime Poco C81x 4G ₹10,075 OnePlus 10R Prime Blue Edition

**Companies:** Intel, Dell Technologies, Lenovo, HPE, Asus, Supermicro
**Countries:** USA

[Read the full story on Times of India](https://timesofindia.indiatimes.com/technology/tech-news/intel-launches-xeon-6-processors-as-ai-moves-from-chatbots-to-agents/articleshow/131448284.cms)

---

Canonical: https://newsio.io/zh-TW/n/c678cac6-6c11-42c7-b3b6-3d4263b82563/computex-ai
Summarized by Newsio from Times of India. https://newsio.io/how-it-works