Intel and SambaNova Announce AI Chip Collaboration
Intel and AI startup SambaNova have announced a multi-year partnership to develop AI inference solutions combining Intel's Xeon CPUs with SambaNova's accelerators. The deal follows reportedly failed acquisition talks and positions the companies to challenge NVIDIA's market dominance. As part of the announcement, SambaNova unveiled its SN50 chip, described as the fastest processor for agentic AI, and announced a new $350 million funding round that included Intel Capital.
- This partnership follows reports from late 2025 that Intel was in advanced negotiations to acquire SambaNova for approximately $1.6 billion, a deal that did not come to fruition. - Intel's CEO, Lip-Bu Tan, also serves as the executive chairman of SambaNova's board and was an early investor in the startup through his venture capital firm, Walden International. - The new SN50 chip is built on a "Reconfigurable Data Unit" (RDU) architecture, a dataflow design that differs from the instruction-set architecture of traditional CPUs and GPUs to minimize data movement bottlenecks during AI processing. - SambaNova's SN50 accelerator is purpose-built for AI inference rather than training, focusing on low latency for real-time applications and featuring a three-tier memory architecture to handle models with over 10 trillion parameters. - The collaboration is aimed squarely at the AI inference market, which analysts believe is more open to competition, in contrast to the AI training market where NVIDIA's market share is estimated to be above 80%. - The SN50 is specifically optimized for "agentic AI," a type of AI that goes beyond content generation to autonomously plan and execute multi-step tasks to achieve a specific goal with limited human supervision. - The new $350 million Series E funding round was co-led by Vista Equity Partners and Cambium Capital, with Vista's participation noted as a rare investment in a silicon company for the private equity firm, which typically focuses on enterprise software. - SoftBank will be the first customer to deploy the SN50 chip, planning to integrate it into its next-generation AI data centers in Japan as part of its sovereign AI initiatives.
