Intel is reportedly in preliminary discussions to acquire U.S.-based AI chip startup SambaNova Systems, a Silicon Valley unicorn co-founded by Chinese-American entrepreneurs and chaired by former Intel CEO and current Walden International chairman Chen Li-fu.According to sources cited by foreign media, Intel is negotiating with SambaNova over potential acquisition terms. The startup has also been working with financial institutions to gauge interest from possible buyers.While talks are still in early stages and no deal is guaranteed — and other buyers could emerge — any agreement would likely value SambaNova below its 2021 valuation of $5 billion .Per a report from The Information, the company may face a significant valuation cut. Secondary market data tracked by Caplight indicates that asset manager BlackRock has marked down the value of its SambaNova holdings by 17% over the past year, bringing the implied valuation down to $2.4 billion — just half of its last funding round.A SambaNova spokesperson stated the company is always exploring strategic opportunities that align with its mission and stakeholders, but declined to comment further. Intel representatives also declined to comment.

Who Is SambaNova?

Founded in 2017 by a group of Stanford University professors, SambaNova has positioned itself as a leading innovator in AI chips and inference systems. The startup offers both hardware and software solutions tailored for cloud-scale AI deployments and generative AI applications.SambaNova’s leadership team boasts impressive academic and industry credentials:

  • •Rodrigo Liang (CEO & Co-founder):A Stanford electrical engineering graduate with over 20 years of experience in semiconductor engineering. He previously held leadership roles at HP, Sun Microsystems, and Oracle, where he led teams building high-performance processors for enterprise systems.
  • •Kunle Olukotun (CTO & Co-founder):Cadence Design Professor of Electrical Engineering and Computer Science at Stanford. He is a pioneer in chip multiprocessing design, founder of Afara Websystems (acquired by Sun in 2002), director of Stanford’s Pervasive Parallelism Lab, and co-leader of the DAWN research project on data analytics.
  • •Christopher Ré (Co-founder):Associate Professor in Stanford’s InfoLab, affiliated with the Statistical Machine Learning Group, Pervasive Parallelism Lab, and Stanford AI Lab. He is also a recipient of the MacArthur “Genius” Fellowship, one of the highest honors for interdisciplinary achievement in the U.S.

Rodrigo Liang recently shared insights on chip design via the TED community, emphasizing SambaNova’s mission to deliver AI chips with the lowest power consumption per rack, optimized for the latest cloud-scale AI models and applications.

Strategic Backing and Previous Valuation

SambaNova’s prestigious founding team attracted high-profile investors early on.

  • •Walden International, led by Chen Li-fu, was one of the startup’s earliest backers, leading its $56 million Series A round in 2018.
  • •In 2021, the SoftBank Vision Fund 2led a massive 676millionfundinground∗∗,joinedbyIntelCapital,GoogleVentures,andWaldenInternational.ThatroundpropelledSambaNova’svaluationto∗∗5 billion, making it one of the most valuable AI chip startups globally at the time.

However, the company’s fortunes have shifted amid broader turbulence in the AI chip sector.

Why Now? Industry Challenges and Market Realities

SambaNova’s reported consideration of a sale reflects the growing pressures facing AI chip startupsin an increasingly competitive and capital-intensive landscape.Earlier this year, in April, the company laid off approximately 15% of its workforce. Its current revenue remains undisclosed.Other AI chip startups are also struggling. For instance, Groq, another U.S. AI chip unicorn with Saudi ties, slashed its 2025 revenue forecast from 2billiontojust500 million.Similarly, Ambarella, a leading U.S. AI vision chipmaker, is reportedly exploring strategic options including a potential sale. Its stock jumped 21%following the news, pushing its valuation to $2.6 billion.The trend of acquisitions targeting AI chip startups is heating up:

  • •NXP Semiconductorsacquired U.S. edge AI startup Kinarafor $307 millionin February.
  • •Metais reportedly eyeing a takeover of Korean AI chip startup FuriosaAI.
  • •AMDacquired the team behind Canadian AI chip startup Untether AIin June.
  • •Metais also said to be pursuing U.S. AI inference chip startup Rivos.

Intel’s Ambitious AI Chip Push

This potential acquisition comes as Intel ramps up its efforts to reclaim leadership in the AI chip space, especially in the data center segment, where Nvidia currently dominates.Intel has a track record of acquiring AI and semiconductor startups:

  • •December 2015:Acquired FPGA giant Alterafor $16.7 billion.
  • •August 2016:Bought AI chip startup Nervana Systemsfor $300–400 million.
  • •March 2017:Acquired Israeli autonomous driving chip company Mobileyefor $15.3 billion.
  • •July 2018:Acquired U.S. structured ASIC provider eASIC.
  • •December 2019:Bought Israeli cloud AI chip startup Habana Labsfor $2 billion.

Now, nearly six years after acquiring Habana Labs, Intel is under pressure to deliver resultsas Nvidia continues to dominate AI training and inference markets.In October, Intel unveiled its new AI execution roadmap, outlining a multi-pronged strategy:

  1. 1.Deliver Agentic AI Infrastructure:
    • •Provide differentiated system-level solutions, including Xeon server CPUs, Gaudi AI chips, Arc GPUs, and AI PCs.
    • •Build an open AI software stackfor seamless, frictionless AI deployment.
  2. 2.Expand Agentic AI Solutions:
    • •Develop inference-enhanced GPUsand foster an open developer ecosystemto grow the overall AI market.
  3. 3.Scale Technologies and Infrastructure:
    • •Advance next-gen inference-optimized GPUs and the Shore product line, customized for Agentic AI and training workloads.
    • •Address bandwidth bottlenecks.

Among its flagship offerings is the upcoming Gaudi 2 (codenamed Jaguar Shores), an AI training chip designed for rack-scale deployments, built on Intel 18A process nodeand equipped with SK Hynix’s HBM4 memory.Sachin Katti, Intel’s Senior Vice President and Chief Technology & AI Officer, stated during a recent tech event:

“Intel is fully committed to deepening its AI capabilities across the entire product portfolio.”

Strategic Implications of a Potential Deal

Acquiring SambaNovacould give Intel a significant boost in its AI chip division — particularly in AI inference, where the startup has focused much of its recent efforts.SambaNova has pivoted toward AI inference applications, offering both cloud-based and on-premises solutionspowered by its proprietary chips and software stack. The company also sees promising growth in the Middle East, where it partnered with Saudi Aramcoto develop a large language model called Metabrain, and pledged $140 million in investmentin Saudi Arabia earlier this year.For Intel, bringing SambaNova in-house could help it:

  • •Accelerate innovation in AI inference chips,
  • •Strengthen its software and systems integration capabilities,
  • •Expand its presence in emerging AI markets, and
  • •Integrate top-tier AI talent and Stanford-caliber research expertise.

Conclusion: A High-Stakes Move in the AI Arms Race

If the deal goes through, it would mark another major move by Intel to reassert itself as a leader in AI hardware — and a strategic effort to close the gap with Nvidia.More broadly, it reflects a shifting landscape where AI chip startups face mounting financial and market pressures, and tech giants are increasingly turning to acquisitionsto secure talent, IP, and market share.For now, all eyes are on whether Intel and SambaNova can strike a deal — and what it could mean for the future of AI infrastructure.