在当前与英特尔的深入谈判中，SambaNova 的企业价值与应偿债务的总额共计约 16 亿美元（IT之家 注：现汇率约合 113.08 亿元人民币），而该公司此前的估值高点是 50 亿美元（现汇率约合 353.37 亿元 人民币）。 IT之家 12 月 15 日消息，彭博社当地时间本月 12 日报道称，英特尔正与 AI 芯片企业 SambaNova 就对 后者的收购交易进行深入谈判。不过后者也已与其他潜在财务投资者签署了意向书，情况仍可能发生变 化。 ▲ SambaNova 的 RDU 芯片 ...

Core Insights - Intel is negotiating to acquire AI startup SambaNova for approximately $5 billion, aiming to rapidly build an end-to-end AI ecosystem and close the gap with leaders like Nvidia [1][5] - If the deal is finalized, it will be the largest and most strategically significant acquisition under the new CEO Pat Gelsinger [1] Company Overview - SambaNova was founded in 2017 and specializes in self-developed RDU (Reconfigurable Dataflow Unit) chips and accompanying software, which significantly enhance efficiency in AI inference scenarios compared to traditional GPU solutions [5] - The company offers an integrated AI platform that includes DataScale rack-level systems and the SambaFlow compiler, with applications in finance, healthcare, and government sectors [5] Strategic Implications - Intel believes that acquiring SambaNova will provide immediate access to a mature hardware and software stack, saving years of research and development time, and strengthening its competitive position in the AI inference market [5] - This acquisition aligns with Intel's recently announced "Crescent Island" GPU roadmap, which prioritizes AI inference as a key direction [5] - By integrating SambaNova's technology with its own manufacturing, channels, and customer resources, Intel plans to launch diverse AI solutions covering cloud, edge, and endpoint, thereby enhancing chip value and improving profit margins [5] Leadership Context - Current CEO Pat Gelsinger's investment firm, Walden International, was an early investor in SambaNova, indicating a long-standing relationship between the two companies [5] - Gelsinger has expressed intentions to accelerate AI transformation through acquisitions, and the resumption of negotiations suggests that internal obstacles may have been resolved [5]