AI networking startup Cornelis Networks announced on September 14, 2026 a $205 million funding round led by IAG Capital Partners, the launch of a new Active Compute Fabric architecture and a strategic collaboration with Qualcomm — a combined package aimed squarely at Nvidia's InfiniBand and NVLink dominance in AI data centers.
Active Compute Fabric: The Network Becomes A Compute Layer
Announced ahead of the AI Infra Summit, Active Compute Fabric embeds programmable compute directly into the network so it can perform AllReduce, gradient compression, KV-cache assembly for disaggregated inference and expert dispatch for MoE models as data moves between accelerators. "AI infrastructure is reaching a point where faster endpoints alone are not enough," said Cornelis CEO Lisa Spelman. Cornelis says pre-production simulations show up to 50% less network traffic.
Qualcomm Joins The Anti-Nvidia Networking Push
Qualcomm — better known for mobile silicon than data-center gear — will share the AI Infra Summit stage with Spelman to unveil a rack-scale AI infrastructure partnership. Both companies argue that networks should be treated as a "first-order design decision" alongside accelerators, rather than passive pipes bolted on afterward. Cornelis CMO Brandon Draeger said the two companies are "in advanced stages of joint technology evaluation focused on rack-scale inference."
Where The $205M Goes: CN5000, CN6000 And Open Standards
Cornelis will use the new capital to scale production of its CN5000 and CN6000 switches and to accelerate deployment of Active Compute Fabric, which is built on open standards including Ethernet, UALink for scale-up and Ultra Ethernet for scale-out. The round positions the Intel spinout as one of the best-funded independent challengers in the AI networking race, alongside stories like Ayar Labs' $150M Series E extension and Meta's MTIA v3 Iris custom silicon program.
Reporting based on coverage from SiliconANGLE, HPCwire, Converge Digest and Forbes.