🇬🇧較早收集於 4h

SambaNova 籌資 3.5 億美元挑戰 Nvidia GPU

SambaNova 籌資 3.5 億美元挑戰 Nvidia GPU
PostLinkedIn
🇬🇧閱讀原文: The Register - AI/ML
#ai-chips#fundingsambanova-rdusambanovarduintelnvidiab200

💡Intel-backed $350M for GPU rival in AI inference—Nvidia challenger emerges

⚡ 30-Second TL;DR

有什麼變化

3.5 億美元融資輪由 Intel 支持

為什麼重要

為 AI 建構者提供潛在 Nvidia 替代方案,降低大規模推理部署的依賴與成本,尤其在 GPU 短缺時。

下一步行動

Contact SambaNova sales to benchmark their 5th-gen RDU against your Nvidia inference workloads.

誰應關注:Developers & AI Engineers

關鍵要點

  • 3.5 億美元融資輪由 Intel 支持
  • 推進第 5 代 RDU 用於 AI 推理
  • 宣稱在速度/成本上超越 Nvidia B200
  • 資料流架構作為 GPU 替代方案

🧠 深度解析

背景與延伸:來自公開資料,非原文內容。引用 5 個來源。

🔑 增強重點摘要

  • SambaNova's SN50 chip achieves up to 895 tokens per second per user on Llama 3.3 70B model, outperforming Nvidia B200's 184 tokens per second per user.[2][3]
  • SoftBank is the first customer deploying SN50 in next-generation AI data centers in Japan for low-latency inference services.[2][3][4]
  • Intel's new CEO Lip-Bu Tan serves as SambaNova's Chairman and early investor via Walden International.[3]
  • The partnership includes multi-year collaboration for hardware-software co-design on Intel Xeon infrastructure, with joint co-selling to enterprise channels.[2][3]
📊 競品分析▸ Show
FeatureSambaNova SN50Nvidia B200
Llama 3.3 70B TPS/user895 [2][3]184 [3]
ArchitectureDataflow RDU [1][2]GPU [2]
Interconnect BW2.2 TB/s bidirectional [2]Not specified

🛠️ 技術深入

  • SN50 is the 5th-gen Reconfigurable Dataflow Unit (RDU) accelerator, optimized for large language and multimodal models on Intel Xeon-based infrastructure.[2][3]
  • Equipped with 2.2 TB/s bidirectional chip-to-chip bandwidth via switched fabric for larger scale-up domains.[2]
  • Supports agentic AI workloads and provides scalable, production-ready inference capacity through integration with Intel's CPUs, accelerators, and networking.[3][4]

🔮 前景展望AI analysis grounded in cited sources

SambaNova SN50 ships to customers in 2026
Company expects to ship SN50 accelerators later this year with SoftBank as first customer.[2]
Intel-SambaNova solutions gain enterprise adoption
Multi-year collaboration includes joint co-selling to Intel's global enterprise, cloud, and partner channels.[3]
Dataflow architecture captures GPU inference market share
Partnership provides GPU alternative for generative AI deployments, leveraging Intel's scale and SambaNova's competitive product.[2]

時間線

2017-10
SambaNova Systems founded by former Google engineers.
2020-01
Raises Series B funding and unveils first RDU generation.
2021-05
Secures $676M Series D funding at $5B valuation.
2023-06
Announces SN40L system and Cardiant AI platform.
2026-02
Raises $350M Series E with Intel; unveils SN50 chip and partnership.
📰

AI 週報

閱讀本週精選 AI 大事摘要 →

👉相關動態

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: The Register - AI/ML

這是摘要,不是原文。去看原站,或訂閱每週簡報。

每週 AI 簡報

每週一封,可隨時退訂。