Velaura AI raises $110M series A at $1B+ valuation to tackle AI’s growing power problem

The AI race is running into a constraint that bigger models and faster chips cannot solve on their own: electricity.

Velaura AI bets that the solution starts inside the chip. The Silicon Valley AI infrastructure startup announced Tuesday that it has raised $110 million in Series A funding at a valuation of more than $1 billion, vaulting the company into unicorn territory as data center operators search for ways to squeeze more computing capacity from limited energy supplies.

Seligman Ventures led the round, joined by new investors Capricorn Investment Group and Prosperity7 Ventures. Existing backers Mayfield, Maverick Silicon, MARA, Premji Invest, Samsung Catalyst Fund, and StepStone Group participated.

The financing will fund development and commercialization of Velaura’s AI compute portfolio, including its recently announced Titan Core silicon platform. The company plans to grow its engineering and customer-facing teams and deepen work with companies building AI infrastructure and physical AI systems.

At the center of Velaura’s pitch is a simple thesis: the AI industry’s next bottleneck may not be access to compute, but access to enough electricity to run it.

Hyperscalers are committing hundreds of billions of dollars to AI infrastructure, yet new data centers increasingly face constraints tied to grid connections, generation capacity, cooling, and the time required to bring new electrical infrastructure online. That changes the economics of the AI race. Getting more useful computation from each watt can become nearly as valuable as adding more accelerators.

Velaura says its Titan Core proprietary digital chip IP and design platform can deliver a 2x to 4x improvement in performance per watt for mathematical operations used by AI accelerators without sacrificing performance. The company says the underlying technology has already been deployed across more than 30 million ASICs.

That production history gives Velaura something many young semiconductor startups lack: evidence that its core engineering can move beyond laboratory demonstrations and into chips manufactured at commercial scale.

From hyperscale data centers to physical AI

Velaura is targeting both ends of the emerging AI compute market.

One is the data center, where electricity consumption has become a major economic and infrastructure issue. The other is physical AI, including robots, drones, autonomous machines, and other systems that must perform increasingly demanding AI workloads under tight energy and thermal limits.

The company says it is already working with leading hyperscalers to incorporate its technology into future XPU roadmaps. If those engagements turn into large-scale deployments, Velaura could find itself sitting inside a part of the AI stack receiving far more attention as infrastructure costs climb.

Its leadership brings considerable semiconductor experience to that effort. Velaura’s executives and engineers have backgrounds at Apple, Nvidia, Google, Qualcomm, and Marvell. The company is led by co-founder and CEO Rajiv Khemani, with a team that includes veteran chip executives who have previously built semiconductor businesses and shipped products at large scale.

“Every advance in AI, from reasoning models to embodied intelligence, creates demand for more compute, and ultimately more power,” Khemani said. “The next era of AI will be defined not only by better models, but also by fundamentally better compute economics. Velaura is building the ultra-low-power silicon and software foundation needed to scale AI from hyperscale data centers to intelligent machines operating in the physical world.”

The $1 billion-plus valuation reflects investor expectations that energy efficiency will become a bigger piece of AI infrastructure economics. It remains an ambitious bet. Semiconductor development is expensive, hyperscaler qualification cycles can be long, and claimed efficiency improvements must hold up across real workloads and production environments.

Yet the timing is hard to ignore. The AI industry spent the past several years asking how many GPUs it could obtain. The next question may increasingly be how many of those chips the electrical grid can afford to keep running.