California-based Upscale AI has raised $200 million in a Series A round, taking the startup’s total funding to more than $300 million since its launch last year, as investors see networking as the next major bottleneck in scaling artificial intelligence.
The financing was led by Tiger Global, India’s Premji Invest, and Temasek-backed Xora Innovation, with participation from Maverick Silicon, StepStone Group, Mayfield, Prosperity7 Ventures, Intel Capital, and Qualcomm Ventures, per an announcement.
As AI models grow larger and more complex, the performance of clusters is increasingly constrained by networking designed for a pre-AI era, executives and investors say. Traditional architectures focus on connecting endpoints such as servers and storage, whereas AI workloads require tightly synchronised systems that function as a single, unified engine.
Upscale AI was founded to address that gap. The company is developing a full-stack, AI-native networking platform that integrates GPUs, AI accelerators, memory, storage, and networking into a synchronised system. Its SkyHamme scale-up solution is designed to enable “unified racks” by collapsing latency between accelerators and other components, allowing AI clusters to scale more efficiently.
Unlike many proprietary approaches in the market, Upscale AI is building its platform on open standards and open-source technologies, including Ultra Accelerator Link, Ultra Ethernet, SONiC, and the Switch Abstraction Interface. The company is an active contributor to industry bodies such as the Ultra Accelerator Link Consortium, Ultra Ethernet Consortium, the Open Compute Project, and the SONiC Foundation.
The fresh capital will be used to expand engineering, sales, and operations as Upscale AI moves towards commercial deployment. The company said it is seeing strong early demand from hyperscalers and AI infrastructure operators seeking open alternatives to closed, vendor-locked networking systems. Its networking solutions are expected to begin shipping later this year.
“This investment accelerates our mission to fundamentally re-architect networking for the AI era,” CEO Barun Kar said, adding that the company sees a “once-in-a-generation opportunity” to build an open AI networking platform as demand for large-scale AI infrastructure surges.
Executive chairman Rajiv Khemani said the pace of adoption reflects mounting pressure on AI operators to overcome networking constraints as compute continues to scale. “The market is demanding open, scalable AI networking solutions, and Upscale AI is uniquely positioned to help customers break through today’s limitations,” he said.
