Upscale AI extends Series A with $190M, doubles valuation to $2B

Upscale AI, a Santa Clara-based startup building networking infrastructure for large-scale AI workloads, has secured a $190 million extension to its early-stage financing. The new capital is structured as a Series A-1 round led by Premji Invest, with participation from both new and existing backers including Nvidia, Salesforce Ventures, Temasek, Seligman Ventures, Mayfield, Tiger Global, StepStone Group, Maverick Silicon, and Prosperity7 Ventures. The extension brings Upscale AI’s total funding to $500 million less than two years after inception and values the company at $2 billion, up from a $1 billion valuation following a previous $200 million Series A funding in January.

The company develops AI-focused networking switch silicon and a full-stack networking architecture designed for GPU-dense data centers, aiming to address bandwidth, latency, and scalability demands of modern AI training and inference clusters. By pursuing an open-standard networking fabric that can connect GPUs from multiple vendors, Upscale AI is positioning itself as an alternative to proprietary data center networking stacks and as a challenger in a market increasingly shaped by hyperscale cloud providers and major chipmakers. The sizable extension round, the presence of strategic investors like Nvidia and Salesforce Ventures, and participation from global institutional investors such as Temasek and Premji Invest underscore growing investor conviction that AI-specific networking infrastructure is a critical bottleneck and a major value layer in the broader AI ecosystem.