Upscale AI hits $2B valuation with $190M Series A extension; Nvidia backs AI networking chip startup

Upscale AI, a Santa Clara-based AI networking infrastructure startup, raised $190 million in Series A-1 extension financing on June 22, bringing total funding to $500 million and valuation to $2 billion. The round was led by Premji Invest and included new backers Nvidia, Salesforce Ventures, Seligman Ventures, and Temasek, alongside existing investors including Tiger Global, Mayfield, and Prosperity7 Ventures. Nvidia's participation as a strategic investor signals confidence in Upscale's architecture for connecting its GPU infrastructure.

Upscale AI builds a full-stack portfolio spanning silicon, systems, and software to solve one of AI's most critical bottlenecks: networking. Modern AI models run across thousands of GPUs that must synchronize at cluster scale; the network connecting them is now a primary limiter on training throughput and inference latency. Upscale designs networking switches and open-standard AI fabric that connects accelerators, memory, and storage into a unified high-performance engine, reducing packet loss and maximizing hardware utilization during large-scale synchronized workloads.

The company raised $200 million in Series A in January 2026, then quickly extended the round by $190 million in June—$500 million in under 18 months from founding. This reflects investor conviction that AI data center networking spending will exceed $100 billion annually by 2030. Upscale's open-standards approach is critical: it reduces vendor lock-in and appeals to hyperscalers and neocloud operators building heterogeneous environments rather than tightly controlled Nvidia-centric stacks.

For architects: Upscale is playing the picks-and-shovels game—the less glamorous infrastructure layer that outlasts individual compute cycles. With Nvidia's validation and customers including hyperscalers, the company is betting that solving the networking bottleneck in large-scale training clusters will become a higher-margin business than GPUs themselves. The open-standards positioning is the key differentiator against entrenched vendors.