Silicom: total expected 2026 AI inference revenues now in multi-million-dollar range

Silicom Ltd. (NASDAQ: SILC) has seen a significant increase in its anticipated 2026 AI inference revenues, with total expected revenues now reaching a multi-million-dollar range. This development follows the selection of Silicom’s high-performance, inference-specific solutions by a pioneering AI infrastructure provider for use in Proof of Concept (PoC) efforts targeting Tier-1 hyperscalers. The PoC is scheduled for the second half of 2026, with initial orders already placed by the customer for delivery in the first half of the year.

If the PoC is successful, the customer anticipates transitioning to a full-scale deployment requiring tens of thousands of Silicom units, each with a multi-thousand-dollar Average Selling Price (ASP). This potential deployment would represent a substantial revenue opportunity for the company. Silicom has also secured orders from two other leading AI compute contenders for distinct inference products, further demonstrating growing market traction.

The company’s inference-specific solutions leverage advanced FPGAs and high-performance COTS networking chips to address challenges such as latency and hardware scalability. Silicom’s CEO, Liron Eizenman, highlighted the industry’s shift from general-purpose GPUs to purpose-built architectures for inference workloads, emphasizing lower total cost of ownership and improved energy efficiency.

These developments underscore Silicom’s position in the rapidly expanding AI inference market and its ability to secure strategic partnerships with key industry players. As the AI inference segment continues to grow, Silicom’s performance in this area could play a pivotal role in shaping its financial trajectory in the coming years.