Intel Fails to Seize AI Opportunity with SambaNova as Acquisition Talks Conclude, Opting for Xeon CPU Collaboration

Intel’s acquisition of SambaNova was initially perceived as a pivotal move to elevate the company’s role in AI inference. However, it seems that Team Blue may have accepted a less ambitious approach than expected.

Intel’s CEO Plans Direct Investment in SambaNova’s Funding Round After Prior Support through Walden Capital

When discussing AI advancements, Intel stands out as one of the few major computing companies that has struggled to seize the opportunities presented by the AI boom. This challenge has lingered for several quarters, and previous CEO Pat Gelsinger has openly recognized the company’s shortcomings in this domain. The decision to overlook the acquisition of SambaNova seems to exemplify this ongoing issue. According to Intel’s recent blog announcement, the company has entered a strategic partnership with SambaNova to develop “Xeon-based”infrastructure along with a co-engineered inference solution that merges Intel and SambaNova features.

This collaboration complements Intel’s existing data center GPU commitments and does not alter its path forward to competing in AI. The company continues to invest across GPU IP, architecture, products, software, systems, and strengthen its roadmap as part of its edge-to-cloud AI engagements.

Together, Intel and SambaNova aim to help shape the future of heterogeneous AI data centers, integrating Intel Xeon processors, Intel GPUs, Intel networking and storage, and SambaNova systems to unlock a multi-billion-dollar inference market opportunity.

The concept behind SambaNova is straightforward yet innovative. This company utilizes specialized processing engines called Reconfigurable Data Units (RDUs) that can effectively translate entire neural network graphs directly into hardware configurations, akin to the approach taken by Etched or Taalas AI. Recently, SambaNova presented its SN50 AI chip, showcasing a remarkable reduction in operational costs—three times lower than GPUs for agentic AI tasks—with a fivefold increase in compute efficiency per accelerator compared to its predecessor. Rather than completing an acquisition, Intel’s revised strategy appears to focus on collaboration to accelerate their entry into the inference market.

SambaNova is in the process of raising $350 million during its Series E funding round, supported by key investors like SoftBank and Intel Capital. Notably, Lip-Bu Tan, an existing investor through his Walden Capital portfolio, is further solidifying his commitment to SambaNova. This dual investment reflects a strong belief in the company’s potential to leverage the inference trend. With the acquisition off the table, what additional steps may Intel take moving forward?

A person holding a semiconductor chip labeled with 'Samantha, ' 'CXERA-IN, ' 'SDR 770FY-35, ' and 'WARDEN 023.' — Image Credits: SambaNova

Analysis indicates that Intel may be gearing up to incorporate SambaNova’s RDUs into targeted workloads, possibly mirroring NVIDIA’s strategy with Groq. Hints of this direction were found in Intel’s latest blog post, which outlines plans to develop “heterogeneous”AI data centers utilizing Intel GPUs, CPUs, and SambaNova systems. As Intel has already faced setbacks in the training sector, the urgency to effectively implement inference solutions cannot be understated, with potential repercussions if they falter in this critical area of AI.

Source & Images