Intel is advancing a key acquisition—purchasing AI inference chip developer SambaNova Systems for $1.6 billion( including debt). Several insiders have revealed that negotiations are in the final stages, and the deal could be finalized as early as January next year. If successful, this will be the first significant corporate acquisition since Intel's current CEO, Pat Gelsinger, joined in March.

**Technological Innovation: Energy-Efficient Inference Chips**

Founded in 2017 by Stanford University professors, SambaNova includes a founder who has received a MacArthur Genius Grant. The company mainly focuses on customized AI chip design, directly competing with Nvidia's market position. Its core product, SN40L, is optimized specifically for AI inference—that is, the phase where models run in production after training.

What makes SN40L competitive? Energy efficiency. Compared to similar products, it can generate more Tokens per kilowatt-hour, which is crucial for controlling data center costs. Its technical principle involves compressing multiple inference computations into a single operation, significantly reducing data transfers between memory and chip cores—this data movement accounts for a large portion of AI chip energy consumption.

SN40L adopts a three-layer memory architecture: high-speed SRAM, HBM, and DRAM, enabling quick loading of AI models and supporting rapid model switching. SambaNova also integrates its chips into SambaRack devices( each containing 16 chips), paired with the custom SambaStack software stack. Additionally, the company offers chip access via cloud services, allowing customers to use the hardware without managing it themselves.

**Funding and Founder's Dual Roles**

Interestingly, Pat Gelsinger currently serves as Chairman of SambaNova. His investment firm, Walden International, is a founding investor in the company, leading a $56 million Series A round in 2018 in partnership with Google’s venture arm( GV). In April 2019, SambaNova raised $150 million in Series B led by Intel Capital; in February 2020, it completed a $250 million Series C led by BlackRock; and in April 2021, it secured $676 million in Series D led by SoftBank Vision Fund 2( with a valuation exceeding $5 billion). Total funding has surpassed $1 billion.

**Market Status: Nvidia Monopoly and Breakthrough Challenges**

Nvidia dominates the AI chip market with approximately 85% market share, establishing an insurmountable moat with its CUDA ecosystem. Other major competitors include Intel and UK-based chipmaker Graphcore. Under Nvidia’s nearly absolute market control, SambaNova faced financing difficulties after 2021, with new funding rounds delayed. Completing this $1.6 billion acquisition allows Intel to acquire a long-term AI chip development platform at a discounted price, aiding rapid expansion of its AI product line.

**Stock Price and Market Reaction**

Since the U.S. government announced in August that it holds a 10% stake in Intel, the company's stock price has surged approximately 60%, rising from before the announcement to recent levels. In the latest trading session, Intel's stock fell slightly by 4.3% to $37.81, with a market capitalization of about $180 billion. Although negotiations are in the final stages, deal terms and timing could still change. Additionally, SambaNova has signed term sheets with other potential investors, and there remains the possibility of pursuing alternative development paths.

**Industry Insights**

Intel’s move reflects the accelerating consolidation in the AI infrastructure market, with major chip manufacturers increasingly investing in high-energy-efficiency inference chips to meet explosive AI deployment demands. This indicates a shift in the value focus of AI chips from general-purpose training chips toward more specialized, energy-efficient inference solutions. In particular, low-power, high-performance solutions for data center scenarios are expected to become the core of future competition.