Upscale AI raises $190M to build open networking for AI data centres
What's the deal? Upscale AI, a Santa Clara-based startup, has raised $190 million in a Series A-1 round, bringing its total funding to $500 million and its valuation to $2 billion — all in under 18 months. Premji Invest, a $15 billion fund, led the round. New investors include Nvidia, Salesforce Ventures, Temasek, and Seligman VenturesDealroom has a profile for this one. Try Dealroom →, with Mayfield, Tiger Global, StepStone, Maverick SiliconDealroom has a profile for this one. Try Dealroom →, and Prosperity7Dealroom has a profile for this one. Try Dealroom → returning.
The company is building open-standard networking fabric that lets AI chips from different manufacturers communicate at full speed. Right now, GPUs powering tools like ChatGPT mostly can't talk to each other efficiently unless they're from the same maker.
"We are trying to get those GPUs to talk the same language," cofounder and chief executive officer Barun Kar told Fortune.
Why now? Spending on AI data centre switches is forecast to surpass $100 billion annually by 2030, according to Dell'Oro Group. The five largest tech companies are expected to spend roughly $660 billion to $690 billion on infrastructure in 2026 alone — nearly double what they spent in 2025.
"Legacy data center networks were designed for a pre-AI world, not for the massive, tightly synchronized scale-up required by modern AI workloads," said Umesh Padval, managing partner at Seligman Ventures.
Every major computing shift has produced a dominant networking company. PCs connecting to the internet gave rise to Cisco and JuniperDealroom has a profile for this one. Try Dealroom →. Cloud computing elevated Arista and Broadcom. Upscale is betting AI is the next such paradigm shift.
What could go wrong? Nvidia's proprietary NVLink and InfiniBand technologies currently dominate AI data centre networking and lock customers into its ecosystem. Upscale must convince hyperscalers that an open-standard approach is worth adopting when Nvidia's integrated stack already works. Notably, Nvidia itself invested in this round — but that doesn't guarantee it won't compete aggressively to protect its networking moat.
The signal: The investment reflects growing demand for interoperability in AI infrastructure. As companies deploy chips from multiple vendors — AMD, Intel, custom silicon from Google and Amazon — the need for vendor-neutral networking grows. A $2 billion valuation in under 18 months signals investors see this as a massive, fast-moving market. If AI infrastructure follows the pattern of previous computing eras, open networking standards could eventually unseat proprietary ones — and Upscale is positioning itself to lead that transition.
