Upscale AI has raised $190 million at a valuation of $2 billion, led by Premji Invest, with participation from Nvidia and others.

According to Mars Finance, AI infrastructure startup Upscale AI announced the completion of a new $190 million funding round, with investors including Nvidia and Salesforce Ventures. This round values the company at $2 billion. This funding round is part of an extension of its Series A round, bringing Upscale AI's total funding to $500 million. The round was led by Premji Invest, with participation from new investors including Temasek and Seligman Ventures, and continued investment from existing shareholders such as Maverick Silicon, Mayfield, Prosperity7 Ventures, StepStone Group, and Tiger Global. Upscale AI focuses on building network infrastructure for large-scale AI clusters, attempting to solve the "computing power interconnect bottleneck" problem in AI training and inference. The company is developing a full-stack AI network architecture covering chips, systems, and software, aiming to build an open standard network interconnect solution suitable for large-scale AI workloads. The company stated that as AI clusters continue to expand, network performance is becoming a key limiting factor affecting GPU utilization and training efficiency; inefficient interconnects directly lead to wasted computing power and increased training costs. Currently, Upscale AI has partnered with several hyperscale cloud vendors and "new cloud" infrastructure providers for evaluation and deployment, but has not disclosed specific customer information. Analysts point out that AI infrastructure investment is spreading from the model layer to the underlying hardware and network layers, with network interconnection becoming one of the most watched key infrastructure tracks after GPUs.