Upscale AI Hits $1 Billion Valuation With $200M Funding to Challenge Nvidia and Broadcom

Upscale AI Secures $200 Million Funding at $1 Billion Valuation

Upscale AI has raised $200 million in Series A funding that values the AI networking startup at more than $1 billion, positioning itself to challenge established players like Nvidia, Cisco, and Broadcom in the critical infrastructure market 1

1

. The round was led by Tiger Global Management, Premji Invest, and Xora Innovation, with participation from Intel Capital, Qualcomm Ventures, Maverick Silicon, StepStone Group, Mayfield, and Prosperity7 Ventures 1

1

. This marks the second major capital injection for the Santa Clara-based company, coming just four months after it announced a $100 million seed round 1

1

.

Tackling Network Congestion and Data Loss in AI Infrastructure

The company aims to solve one of AI's most pressing bottlenecks: network congestion and data loss in cloud-computing systems that train and run AI programs. "AI networking is one of the most urgent and unaddressed bottlenecks in the system," said Barun Kar, Upscale AI's chief executive officer. "A faster chip cannot just deliver breakthroughs if the network can't keep up" 1

1

.

While Cisco, Broadcom, and Arista Networks are addressing similar challenges, Upscale AI seeks an edge by building purpose-built switches for AI rather than adapting legacy technology 1

1

.

SkyHammer Networking Chip Targets Nvidia's NVSwitch Dominance

At the heart of Upscale AI's strategy is SkyHammer, a custom ASIC designed specifically for scale-up AI networking to compete directly with NVSwitch chips found in Nvidia's NVL72 racks 2

2

. The SkyHammer networking chip will offer deterministic latency, enabling administrators to predict data transmission times with high certainty and avoid unexpected delays that slow AI workloads 3

3

. "Instead of retrofitting legacy systems, what we are re-imagining is what scale truly means in AI networking," Kar explained 2

2

.

The chip will feature acceleration for collective communication similar to Nvidia's Sharp and support multiple high-speed interconnect protocols including UALink and ESUN 2

2

. ESUN, which launched last year with backing from Nvidia, Broadcom, and other major players, is designed specifically for scale-up networking inside racks 3

3

. SkyHammer will also support UEC, a technology geared toward connecting different racks that can power AI clusters with up to 1 million chips 3

3

.

Expansion Plans and Market Validation

Upscale AI will use the fresh capital to expand its engineering operation from 150 employees to approximately 300 by year-end 1

1

. The company is already in discussions with data center giants, specialized cloud-computing providers, and sovereign AI projects backed by governments 1

1

. "We are already partnering with hyperscalers and GPU vendors who have validated the architecture. That is complete. Now the focus with this funding is to turn innovation into deployment," Kar stated .

The startup plans to offer SkyHammer as both a standalone ASIC for hyperscale integration and as an integrated switch blade and rack, with products expected later this year 2

2

. To ensure manageability at scale, Upscale AI is extending support for SONiC, the open-source network operating system originally developed by Microsoft that is widely deployed among hyperscale customers 2

2

.

Veteran Leadership and Investor Confidence

The company benefits from experienced leadership. Executive Chairman Rajiv Khemani co-founded and ran Innovium, which was acquired by Marvell Technology Inc., and served on the leadership team at Cavium, another Marvell acquisition 1

1

. Kar was an early employee at Palo Alto Networks and served as senior vice president of engineering 1

1

. Upscale AI was spun out of Auradine, founded in 2022 by Kar and Khemani, which raised more than $300 million from some of the same investors 1

1

.

Navin Chaddha, managing partner at Mayfield, which co-led the earlier round, noted the fundraising happened quickly because investors reached out to the company. "You can solve a problem, but if it's not the No. 1 problem for me, you're wasting my time," Chaddha said, emphasizing that Mayfield seeks companies addressing customers' biggest pain points 1

1

. Premji Invest managing partner Sandesh Patnam added, "As AI systems scale, interconnect efficiency has become a defining driver of performance and economics -- not just raw compute" 3

3

.