SetupAI
EN
來源較早收集於 50m

Microsoft 發表搭載 RTX Spark 晶片的 Surface Laptop Ultra

PostLinkedIn
💻閱讀原文： ZDNet AI
#hardware#edge-ai#computex-2026surface-laptop-ultramicrosoftnvidiasurface laptop ultrartx spark

💡筆電具備 128GB 統一記憶體，將徹底改變本地 LLM 開發與裝置端 AI 推論的遊戲規則。

⚡ 30 秒速覽

有什麼變化

搭載全新 Nvidia RTX Spark 晶片，專為高效能 AI 任務設計

為什麼重要

筆電規格中納入 128GB 統一記憶體，大幅降低了在本地執行大型語言模型（LLM）的門檻。此硬體升級使開發者無需完全依賴雲端基礎設施，即可進行複雜模型的原型設計與測試。

下一步行動

評估您的本地開發工作流程，確認 128GB 統一記憶體是否能讓您將特定的模型微調任務從雲端 GPU 轉移至本地硬體執行。

誰應關注：Developers & AI Engineers

關鍵要點

  • •搭載全新 Nvidia RTX Spark 晶片，專為高效能 AI 任務設計
  • •支援高達 128GB 統一記憶體，適合大型模型本地推論
  • •於 Computex 2026 發表，成為 Surface 系列全新旗艦產品

🧠 深度解析

背景與延伸：來自公開資料，非原文內容。引用 16 個來源。

🔑 增強重點摘要

  • •The Surface Laptop Ultra is powered by Nvidia's RTX Spark SoC, which is an Arm-based chip co-developed with MediaTek and manufactured on TSMC's 3nm process.
  • •The RTX Spark chip integrates a 20-core Nvidia Grace CPU with a Blackwell-based GPU featuring 6,144 CUDA cores and fifth-generation Tensor cores, delivering up to 1 petaflop of FP4 AI performance.
  • •The device is designed to run large AI models up to 120 billion parameters locally, leveraging its 128GB of unified LPDDR5X memory with 300 GB/s bandwidth, dynamically allocated between the CPU and GPU via NVLink C2C.
  • •Microsoft positions the Surface Laptop Ultra as a direct competitor to the Apple MacBook Pro, targeting AI development, creative professionals, and developers with its high-performance capabilities and a 15-inch mini-LED PixelSense Ultra display offering 2,000 nits peak HDR brightness.
  • •The Surface Laptop Ultra features a robust dual-fan, dual heat-pipe cooling system, providing more than twice the thermal capacity of the previous Surface Laptop 7th Edition, to sustain its high performance.
📊 競品分析▸ Show
Feature/ProductMicrosoft Surface Laptop UltraApple MacBook Pro (M5 Pro)Dell XPS 16 (2026)Asus ROG Flow Z13 (2026)
ProcessorNvidia RTX Spark (20-core Grace CPU, Blackwell GPU)Apple M5 Pro (Arm-based)Intel Panther Lake (Core Ultra X7 358H)AMD Ryzen AI Max+ 395
AI PerformanceUp to 1 Petaflop (FP4)Not specified (M5 Pro)High TOPs for LLMsRuns 120B parameter AI model
Unified MemoryUp to 128GB LPDDR5XNot specified (M5 Pro)Not specifiedNot specified
Display15-inch mini-LED PixelSense Ultra (2880x1920, 2000 nits HDR)Not specified16-inch Tandem OLED touch display13-inch 2.5K Mini-LED
Target AudienceAI development, creators, developersCreators, professionalsGeneral AI workloads, professionalsLocalized AI, portability
Operating SystemWindows on ArmmacOSWindowsWindows
AvailabilityFall 2026Not specified (2026 model)20262026
Starting PriceUnannounced (expected premium)Not specifiedNot specifiedNot specified

🛠️ 技術深入

  • SoC Architecture: Nvidia RTX Spark, based on Grace Blackwell architecture, built on TSMC's 3nm process.
  • CPU: 20-core Nvidia Grace CPU (Arm-based), co-developed with MediaTek.
  • GPU: Blackwell-based integrated GPU with 6,144 CUDA cores and fifth-generation Tensor Cores for on-device AI processing.
  • AI Compute Performance: Up to 1 petaflop of FP4 AI performance.
  • Memory: Up to 128GB LPDDR5X unified memory with 300 GB/s bandwidth, dynamically allocated between CPU and GPU.
  • Interconnect: NVLink C2C (chip-to-chip interconnect) connects the CPU and GPU.
  • Local AI Capabilities: Capable of running 120-billion-parameter AI models locally.
  • Display Technology: 15-inch mini-LED PixelSense Ultra touchscreen display with 2880 x 1920 resolution (262 ppi) and up to 2,000 nits peak HDR brightness.
  • Thermal Management: Optimized internal layout with dual-fans and dual heat-pipes, offering more than twice the thermal capacity of the Surface Laptop 7th Edition.
  • Software Stack: Supports the full Nvidia AI and graphics technology stack, including CUDA, Tensor RT, NVFP4, DLSS 4.5, RTX ray tracing, Reflex, and G-SYNC.
  • Windows on Arm Optimizations: Includes optimizations for unified memory, heterogeneous scheduling, power and thermal management, and the Windows 11 Prism emulator for x86/x86-64 applications.
  • Security: Designed with OpenShell runtime enhanced with Microsoft-designed security and containment primitives, supporting Hermes and OpenClaw AI agent frameworks.

🔮 前景展望基於引用來源的 AI 分析

The Surface Laptop Ultra will significantly accelerate the adoption of Windows on Arm for high-performance computing.
Its powerful Nvidia RTX Spark chip, unified memory architecture, and Microsoft's flagship endorsement directly challenge traditional x86 dominance and Apple's M-series, making Windows on Arm a viable option for demanding professional workloads.
The integration of 1 petaflop AI compute and 128GB unified memory will drive a new era of on-device AI agent development and deployment.
This level of local processing power enables complex AI models (120B+ parameters) and agent frameworks to run securely and efficiently on a personal device, reducing reliance on cloud infrastructure and enhancing privacy.
The Surface Laptop Ultra will set a new benchmark for premium AI PC design, influencing future laptop form factors and thermal solutions.
Microsoft's focus on a traditional, yet thermally optimized design with a high-brightness mini-LED display and robust port selection, combined with Nvidia's powerful SoC, demonstrates a shift towards balancing extreme performance with practical portability.

⏳ 時間線

2012-06
Microsoft announces first-generation Surface devices, marking its entry into branded PC hardware.
2015-10
Microsoft introduces the Surface Book, a hybrid laptop with dedicated graphics, targeting creative professionals.
2019-10
Surface Pro X debuts with an ARM-based processor, reintroducing Arm to the Surface family for improved battery life and connectivity.
2024-05
Microsoft launches Copilot+ PCs with Qualcomm Snapdragon processors, emphasizing AI integration and a renewed focus on Windows on Arm.
2025-01
Reports emerge that Nvidia is developing Windows on Arm chips (codenamed N1 and N1X) in collaboration with MediaTek.
2026-06
Microsoft unveils the Surface Laptop Ultra at Computex 2026, powered by the new Nvidia RTX Spark chip, as its flagship high-performance AI device.
💻閱讀原文： ZDNet AI
📰

AI 週報

閱讀本週精選 AI 大事摘要 →

👉相關動態

同主題

探索 #hardware

同產品

更多關於 surface-laptop-ultra

同來源

最新來自 ZDNet AI

AI 策展新聞聚合。所有內容版權歸原始發布者所有。
原始來源： ZDNet AI ↗

這是摘要，不是原文。去看原站，或訂閱每週簡報。

每週電子報

每週一封，可隨時退訂。