SambaNova has secured $1 billion in Series F funding at an $11 billion valuation, backed by General Atlantic and other major investors, as the company positions itself to solve a critical bottleneck in AI infrastructure: the growing mismatch between what existing data centers can power and what next-generation AI systems demand. The funding announcement comes alongside a stark reality facing the industry: fewer than 8% of enterprise data centers will meet the power and cooling requirements needed to run advanced AI chips when they ship, forcing companies to choose between building expensive new facilities or finding ways to run AI on existing infrastructure.

Why Are Data Centers Struggling to Power Next-Generation AI?

The AI data center power crisis boils down to a simple problem: the systems are getting hungrier for electricity much faster than data centers can adapt. Most enterprise data centers built over the past two decades were designed to support individual server racks consuming just 5 to 8 kilowatts of power, with air cooling as the standard approach. But modern GPU-based AI systems have shattered those assumptions.

NVIDIA's latest Blackwell DGX GB200 NVL72 systems draw up to 120 kilowatts per rack, a more than 10-fold increase from earlier generations. The even newer Vera Rubin NVL72 racks operate at 120 to 130 kilowatts, while the Rubin Ultra NVL576 skyrockets to as much as 600 kilowatts per rack, enough electricity to power 400 homes. Looking further ahead, systems expected to follow Rubin may require as much as 1 megawatt per rack. Meanwhile, existing data centers can typically only support 30 kilowatts or less per air-cooled rack.

The scale of this mismatch is staggering. Gartner estimates that meeting the incremental power needs of AI data centers in 2027 alone will require 500 terawatt-hours of electricity per year, a 2.6-fold increase from 2023 levels and nearly as much as Germany's entire annual power consumption.

What Makes Building New Data Centers So Expensive and Slow?

Rather than upgrade existing facilities, many companies are planning entirely new data centers. But this approach carries enormous costs and timelines. Foxconn estimates that a single 1-gigawatt data center built to power NVIDIA's latest systems would cost approximately $47 billion to construct, with an annual electric bill of roughly $1.3 billion. NVIDIA CEO Jensen Huang has confirmed that building a data center from breaking ground to operational AI supercomputer takes about three years in the United States.

The construction boom is real: Pew Research counts more than 1,500 new data centers currently in various stages of development in the U.S. alone. However, construction timelines are outpacing power infrastructure expansion. One report estimates that the inability to secure adequate power will delay or cancel 30% to 50% of AI data centers planned for deployment in 2026. Of the 16 gigawatts of capacity slated to come online in 2026, only 5 gigawatts are actively under construction.

How Can Existing Data Centers Run Next-Generation AI?

This is where SambaNova's strategy diverges from the broader industry trend. Rather than requiring new construction, the company's RDU (Reconfigurable Data Unit) platform is designed to operate within the power and cooling constraints of existing data centers. SambaNova's SambaRack SN40-16 runs at approximately 10 kilowatts per rack, while the newer SambaRack SN50 operates at 20 kilowatts per rack, both using standard air cooling.

This positioning addresses a critical market need. Existing data centers hold more than 80% of available AI capacity but support only 30 kilowatts or less per air-cooled rack. By offering systems that fit within those constraints, SambaNova enables enterprises to deploy advanced AI without waiting years for new facilities to be built or spending billions on construction.

• Power Efficiency: SambaNova's RDU platform consumes 10 to 20 kilowatts per rack compared to 120 to 600 kilowatts for the latest GPU-based systems, allowing deployment in existing data centers without infrastructure upgrades.
• Sovereign AI Deployment: Nations with limited data center capacity and aging power grids can deploy AI systems without waiting for new construction, addressing national AI initiatives in resource-constrained regions.
• Cost Avoidance: Enterprises can avoid the $47 billion price tag and three-year timeline associated with building new 1-gigawatt data centers, instead leveraging existing infrastructure.

Who Is Backing SambaNova's Vision?

The $1 billion Series F round was led by General Atlantic, with significant participation from Seligman Ventures, T. Rowe Price Associates, and Capital Group. Additional investors include Intel Capital, BlackRock, Qatar Investment Authority, and Vista Equity Partners, among others.

"SambaNova's platform is differentiated, built for a market where inference has become foundational to enterprise and industry transformation," said Martín Escobari, Co-President and Head of Global Growth Equity at General Atlantic. "Rodrigo and the team are driving deep technical innovation to achieve growing commercial momentum while demand for inference is accelerating well ahead of supply."

Martín Escobari, Co-President and Head of Global Growth Equity at General Atlantic

The funding announcement also highlighted a major customer win: JPMorganChase has selected SambaNova as an inference infrastructure partner, deploying its SN40 and SN50 systems to power secure, on-premises AI inference for the financial services firm.

"At JPMorganChase, AI infrastructure has to meet a very high bar for performance, control and reliability. We're excited to deploy SambaNova's RDU architecture and looking forward to testing its speed and security for on-prem inference in our demanding enterprise AI workloads," said Darrin Alves, Chief Information Officer of Infrastructure Platforms at JPMorganChase.

Darrin Alves, Chief Information Officer of Infrastructure Platforms at JPMorganChase

What Does This Mean for the Broader AI Infrastructure Market?

SambaNova's $11 billion valuation reflects investor confidence that inference, not just training, has become central to enterprise AI strategy. As AI moves from experimental projects to production deployment, the focus has shifted from raw training power to efficient, cost-effective inference, where models run on real-world data to generate predictions or responses.

The company plans to use the capital to expand manufacturing capacity, accelerate product innovation, and scale deployments across enterprises, neo-clouds, sovereign AI customers, and service providers worldwide. It will continue investing in chips, systems, software, and full-stack AI infrastructure.

The timing is critical. Teradata estimates that by the time NVIDIA's Vera Rubin systems ship, fewer than 8% of enterprise data centers will have both the power capacity and liquid cooling infrastructure required to run them. For the vast majority of organizations, SambaNova's approach of fitting advanced AI into existing facilities represents a more practical path forward than waiting for new data center construction to complete.

" }