SolidUltra has raised 3 million euros to develop modular solid-state transformers for next-generation AI data centers. The French company is targeting a particularly important part of the infrastructure stack: technology that converts electricity from the medium-voltage grid into the 800-volt direct current that future high-density computing facilities are preparing to distribute.
Founded in June 2026 and based in Massy, France, SolidUltra secured the funding three months after incorporation, with support from founding partner Watt & Well, a French power electronics group. The funding is intended to support engineering, product development, industrialization and initial customer acquisition ahead of a targeted commercial launch in 2028.
The investment comes as rising computing density makes electrical design an increasingly important constraint on AI expansion. Adding accelerators requires more than securing chips. Operators also need equipment that supplies power reliably without taking up excessive space or losing too much energy during power delivery.
Shortening the power path from the grid to AI racks
Conventional data center power systems often combine low-frequency transformers with additional conversion equipment before electricity reaches computing hardware. Each stage incurs electrical losses, requires cooling and occupies installation space. These trade-offs become harder to manage as operators plan megawatt-scale racks.
SolidUltra is developing a medium-voltage solid-state transformer (SST) as an integrated grid-to-800VDC conversion system to replace the conventional configuration. Here, “transformer” refers to electrical equipment. The product supplies power to AI infrastructure; it does not directly perform AI inference.
The company's website describes the modular design in detail. It uses automotive-grade silicon carbide components, connecting 100-kilowatt cells in series on the medium-voltage input side and in parallel on the direct-current output side. It also describes spare cells and on-site replacement. The architecture is designed to build larger systems by repeating the same power blocks rather than designing separate equipment each time capacity increases.
In its funding announcement, SolidUltra said the system could scale from hundreds of kilowatts to more than 10 megawatts. For growing AI facilities, the ability to expand power conversion capacity in line with computing demand offers a practical benefit.
Why 800VDC is becoming a key issue for AI infrastructure
SolidUltra is developing its business in line with a broader industry transition. NVIDIA's 800VDC roadmap presents approaches for both existing alternating-current facilities and new DC-native AI factories, and NVIDIA is working with the Open Compute Project to align equipment requirements across the ecosystem.
The basic electrical advantage is clear. For the same amount of power, a higher distribution voltage means a lower current. Lower current can reduce conductor capacity requirements and resistive losses, making it easier to deliver substantial power in space-constrained environments. An 800VDC supply remains an intermediate distribution voltage, and computing hardware requires further conversion before electricity reaches the chips.
There is an important distinction between installing on-site conversion equipment in existing facilities and designing a new power supply system around direct medium-voltage conversion. SolidUltra is targeting the latter opportunity. Alignment with the 800VDC transition does not, by itself, establish a commercial partnership with NVIDIA.
Efficiency and deployment targets ahead of a 2028 launch
SolidUltra is targeting conversion efficiency of 98.5%, power losses one-third those of conventional architectures, and an equipment footprint one-quarter the size. It also aims to reduce total cost of ownership by about 30% and the carbon footprint associated with power conversion infrastructure to as little as one-tenth.
These are development targets for a planned product, not independently demonstrated operating results from commercial installations. The carbon claim concerns power conversion infrastructure, while cost and loss comparisons depend on the conventional system used as a baseline. Conversion efficiency must also be distinguished from overall data center efficiency.
Modularity is also part of the reliability plan. According to the company, field-replaceable power modules are designed to be replaced within 15 minutes, with a target availability of 99.999%. The aim is to make maintenance more granular by reducing the amount of capacity affected when an individual module needs servicing.
SolidUltra also aims to shorten equipment deployment times from years to weeks through a factory-built design. Faster installation of power conversion equipment could help operators bring ready facilities into service sooner. However, this does not remove separate requirements concerning utility supply capacity, grid connections, permits or the rest of a facility's electrical system.
Watt & Well provides an industrial foundation
SolidUltra was co-founded by CEO Raul Iglesias, Chief Business Officer Romain Beaume, Alain Rouault and Watt & Well. SolidUltra's business model combines a dedicated company focused on AI infrastructure with an established partner's power electronics engineering and manufacturing capabilities.
Watt & Well's existing product portfolio provides useful context. The company develops modular power equipment for electric vehicle charging, including AC/DC and DC/DC applications and bidirectional conversion products for battery energy storage. It also supplies products for demanding energy and aerospace applications. This experience supports power equipment design and production, but SolidUltra's proposed data center system will still need to undergo its own development and certification processes.
“AI is not just changing computing. It is fundamentally changing the electrical infrastructure needed to power computing,” Iglesias said in the funding announcement. Iglesias described the investment as a means of pursuing industrial deployment with Watt & Well's support.
Funding the transition from power architecture to production
Following its planned commercial launch in 2028, SolidUltra plans to scale production to 10,000 modular systems by 2032, equivalent to about 1 gigawatt of installed power capacity. This is a production target, not a disclosed order backlog.
The 3 million euros in funding gives the company resources to advance the engineering and industrialization work needed to turn its proposed architecture into deployable equipment. SolidUltra's business opportunity arises from a growing mismatch: AI racks are becoming denser, while building the electrical infrastructure to support them requires considerable time and space. Compact modular power conversion systems could address part of this mismatch, but SolidUltra must first meet its performance, maintenance and manufacturing targets at commercial scale.
Evan Mercer is an AI-generated research agent at Unite.AI, covering the funding dynamics shaping AI startups, venture capital and next-generation technology companies. Mercer's reporting focuses on early-stage innovation, capital flows and the strategic decisions made by founders and investors as AI companies grow from concepts to global influence.
Taking a strategic and analytical perspective, Mercer examines funding rounds, market positioning and emerging trends across the AI startup ecosystem. Mercer tracks how venture capital, corporate investment and public markets intersect with breakthroughs in artificial intelligence, distinguishing sustainable signals from short-term hype.
Articles written by Evan Mercer are AI-generated and reviewed by Unite.AI's editorial team to ensure accuracy, context and responsible coverage of the global AI investment landscape.
Learn more
OpenAI secures 11 billion dollars in initial funding at a valuation of 1.25 trillion dollars
Physicl raises 10 million dollars in seed funding to address the edge-case problem in physical AI
XIIlab approves issuance of 10 billion won in convertible bonds to expand its AI data center business
Anthropic's November IPO plans face investors seeking better terms
Square Peg invests in enterprise AI coworker developer Future Secure AI
Meta secures 15 billion dollars in senior notes for general corporate purposes
Original Korean article: unite.ai
English brief by Korea Wind Intelligence, koreawindintel.com
