Velaura Raises $110mn for Low-Power AI Chips That Cut Bills, Not Speed

An AI chip startup based in Santa Clara, announced a Series A funding round reported by The Next Web on Wednesday, August 19, 2026, raising $110 million.

The company’s pitch is direct: sell AI chips meant to reduce data-centre power bills without reducing clock speed, and build software around that silicon. Here’s the takeaway: is betting that the next constraint for AI infrastructure is electricity, not performance targets.

If power delivery stays slower than compute demand, hyperscalers will keep paying for expensive capacity—so “efficiency at the same speed” becomes a product requirement, not a nice-to-have.

Velaura: Key Details From ’s Series A and Its Chip Strategy

’s Series A round totals $110 million, and Reuters has reportedly placed the valuation at more than $1 billion, according to coverage of the announcement by The Next Web. The round was led by Seligman Ventures, with participation from Capricorn Investment Group and Prosperity7 Ventures.

Existing backers also joined, including Mayfield, Maverick Silicon, MARA, Premji Invest, the Samsung Catalyst Fund, and StepStone Group. said the funds will go toward speeding development of its chip products and hiring additional engineering and customer-facing staff.

That matters because chip roadmaps are long, and early customer integration often determines whether an efficiency claim survives contact with real workloads. The company’s argument centers on a platform rather than only a standalone accelerator.

Its main offering includes a chip-design platform called Titan Core, paired with ultra-low-power silicon and software targeted at data centres and what the industry calls “physical AI.”

Why Is Framing AI as a Power-Delivery Problem

’s framing is blunt: AI hits a “power wall” before it hits a compute wall. Demand for compute continues rising, but the electricity and infrastructure to support it do not arrive at the same pace.

That mismatch is exactly the kind of bottleneck that turns energy efficiency into a purchasing decision tied to deployment timelines. Hyperscalers, argues, are spending hundreds of billions of dollars to build AI data centres while still waiting for power to come online. For more detail, see VentureBeat AI.

The practical effect is that even when GPUs and accelerators can scale, the facility-level power budget can limit how quickly fleets expand.

OpenAI is cited as an example of how seriously some buyers treat power constraints, including by hiring a power trader so electricity can be managed like a position on its books.

In that context, ’s approach is to reduce power draw while preserving clock speed targets—so customers can keep performance while lowering the electricity cost per workload.

What Comes Next for ’s Titan Core Platform

The next step for is execution: translate an “efficiency without speed loss” promise into production-ready chip products and integration support.

In a market where power planners and workload engineers both influence purchasing, the company will likely need strong early reference designs and measurable outcomes inside real data-centre deployments. That also explains the hiring emphasis in the Series A plan.

Customer-facing engineers help translate architecture claims into system behavior—timing, thermal envelopes, and operational stability—especially when customers are juggling procurement cycles and grid constraints.

For readers tracking the broader AI infrastructure buildout, this is the moment to pay attention to the power layer across the stack, not just model performance.

is now competing for that attention with a silicon-and-software platform built around reducing the electricity bill without changing clock-speed expectations, and the market will judge it as soon as early adopters deploy it.

Related Articles

• SpaceXAI Grok 4.6 (Aug 2026): 500K-Context AI Power
• EVM's EnTwins Speaker Splits in Two So Your Sound Doesn't Have To Stay Boxed In
• No More Guesswork: Embark's New Tool Tells Companies Exactly What a GCC in India Will Cost

FAQs

How much did raise?

raised $110 million in a Series A round, reported by The Next Web on Wednesday, August 19, 2026.

What is selling besides chips?

The main product includes a chip-design platform called Titan Core, alongside ultra-low-power silicon and software for data centres and physical AI.

Does claim lower power without lowering clock speed?

Yes. The company said its chips are designed to reduce power bills without sacrificing clock speed.

Who led ’s Series A?

Seligman Ventures led the Series A, with new investors including Capricorn Investment Group and Prosperity7 Ventures and participation from existing backers.

Closing Takeaway

If AI power delivery keeps lagging demand, ’s “cut the watts, keep the speed” strategy could become a procurement standard rather than an efficiency bonus. Stay tuned for more on Velaura.