Velaura AI Raises $110M for Low-Power AI Compute

Velaura AI has raised $110M in Series A financing at a valuation of more than $1B. Seligman Ventures led the round, with new investor Capricorn Investment Group and returning backers Mayfield, Maverick Silicon, MARA, Premji Invest, Samsung Catalyst Fund, and StepStone Group.

The Santa Clara company develops ultra-low-power silicon, design IP, and software for AI data centers and physical AI systems such as robotics and autonomous machines. Its pitch is direct: AI infrastructure does not only need more compute; it needs better economics for every watt that compute consumes.

That matters because the power problem is no longer hiding in a facilities spreadsheet. The International Energy Agency expects global data-center electricity demand to roughly double from 485 TWh in 2025 to 950 TWh in 2030, while electricity consumption from AI-focused data centers is projected to triple.

What Happened

Velaura announced the financing on August 18, 2026, saying the capital will accelerate development and commercialization of its AI compute portfolio. The company also plans to expand its engineering and customer-facing teams and deepen work with strategic partners and customers building data-center and physical AI systems.

The official announcement identifies Seligman Ventures as lead investor and Capricorn Investment Group as a new participant. Existing investors Mayfield, Maverick Silicon, MARA, Premji Invest, Samsung Catalyst Fund, and StepStone Group also joined the round, while Reuters independently confirmed the $110M amount, Series A classification, valuation threshold, and core investor group.

Velaura previously operated as Auradine and adopted its current identity in March 2026 as it expanded its focus from energy-efficient semiconductor systems into cloud, edge, and physical AI compute. The company and Reuters describe this financing as a Series A, although historical databases carry earlier Auradine funding under other round labels, so a reconciled cumulative-funding total is not used here.

Why Power Efficiency Is the Product

Velaura is attacking AI's electricity problem inside the chip. Its Titan Core platform combines proprietary digital chip IP, low-voltage libraries, and design tools intended to reduce the power used by mathematical operations in AI accelerators while preserving performance.

The company reports a 2-4x improvement in performance per watt for those operations and says the underlying technology has been deployed in more than 30M production ASICs. Those are company-reported figures rather than independent benchmark results, but they show the argument Velaura is taking to accelerator designers: efficiency can be designed into silicon before operators spend more money on power generation, grid connections, and cooling.

Velaura's customer model sharpens that argument. CEO Rajiv Khemani told Reuters that the company charges an upfront technology fee plus a royalty tied to a share of the power savings achieved by customers, a structure similar to an IP licensing model but anchored to energy economics. When efficiency becomes measurable revenue for the supplier and measurable savings for the buyer, performance per watt stops being a technical footnote and becomes part of the commercial contract.

Investor Logic and Market Timing

The investor mix matches the problem. Seligman Ventures focuses on AI infrastructure, cloud infrastructure, next-generation data centers, and cybersecurity, while Capricorn's Technology Impact Funds back deep technologies aimed at climate and resource constraints. Existing investors add semiconductor, growth-capital, and strategic operating experience to a company trying to sell into long, demanding infrastructure cycles.

Reuters reported that Velaura is engaged with 3 of the 4 largest cloud providers as potential customers, though Rajiv Khemani declined to name them. The language matters: engagement is not a commercial deployment, but it indicates the company is testing its proposition with organizations that directly manage the power, cooling, and capital consequences of AI scale.

The broader market is giving those buyers fewer easy choices. The IEA reported that data-center electricity consumption rose 17% in 2025 and expects AI-focused facilities to consume power even faster through 2030, while constraints in grids, transformers, generation equipment, and advanced chip supply are slowing projects. More compute remains the demand signal, but power availability is becoming the speed limit.

The Team Behind the Bet

Velaura's current leadership includes co-founder and CEO Rajiv Khemani, co-founder and Chief Development Officer Manu Gulati, co-founder and President, Strategy & GTM Sanjay Gupta, and President, Products YJ Kim. The company says its team includes veterans of Apple, NVIDIA, Google, Qualcomm, and Marvell, bringing experience across chip architecture, systems, software, product, and go-to-market execution.

Manu Gulati adds a particularly relevant design pedigree after working across Apple, Google, Qualcomm, and NUVIA, which he co-founded. Velaura does not currently list a CTO on its official team page, and advisor Aditya Grover is CTO of Inception AI rather than Velaura, an important distinction in a sector where recycled biographies can turn into bad data surprisingly fast.

The new round gives this team more capital, but the real test is not hiring volume or valuation theater. Velaura must convert company-reported efficiency gains into validated customer outcomes, move potential hyperscaler engagements toward commercial programs, and prove its royalty model can survive the procurement and integration realities of advanced semiconductor design.

What This Signals for AI Infrastructure

The $110M Series A signals that investors are widening the AI infrastructure thesis beyond accelerators, model labs, and data-center construction. Power-aware silicon IP can become a valuable control point if it increases useful compute inside fixed electrical and thermal budgets, especially as new grid capacity and large facilities take years to deliver.

That does not make Velaura's outcome automatic. Chip-design promises must survive verification, manufacturing, software integration, customer qualification, and brutally practical deployment schedules, while the company's largest performance claims still come from its own materials. The financing buys time and engineering capacity, not immunity from semiconductor physics or enterprise procurement.

Still, Velaura is aiming at the right layer of the stack. AI's next infrastructure winners may not be the companies that merely add more megawatts; they may be the ones that make each megawatt carry more intelligence, more economically, across data centers and machines operating in the physical world.