siliXon wants you to design circuit boards by describing them. It just raised $1.5M.
A UK startup co-founded by Romanian engineer Mihai Mesteru turns a sentence into a production-ready PCB. The pre-seed is in; the hard part is whether the boards actually work.
Type "Bluetooth drone controller" into a box and get back a circuit board you can send to a factory. That's the pitch behind siliXon, a UK-based startup co-founded by Romanian engineer Mihai Mesteru, which has just raised more than $1.5 million in pre-seed funding led by German investor System.One.
The promise is blunt: PCB design used to mean months of learning niche EDA software and a hardware engineer's intuition. siliXon says you describe what you want, and its AI returns a manufacturable printed-circuit-board design in minutes.
If that holds up, it's a real wedge. If.
What it actually does
A printed circuit board is the unglamorous substrate under every gadget — drones, smart devices, industrial controllers. Designing one is a specialist craft: pick components, draw the schematic, lay out the copper, and survive a long list of physical constraints before a fab will touch it.
siliXon's platform takes a plain-language product description and generates the schematic and layout end to end. Mesteru — who's hiring, in his words, anyone who loves "eating cables" — is steering the company alongside co-founders Bach Nguyen and Adam French. An open beta is on the way, and the round is earmarked for the usual pre-seed trifecta: build the product, grow the team, get it in front of users.
No customer numbers, no revenue, no adoption metrics yet. At pre-seed, that's expected — you're funding a bet, not a track record.
The part worth being skeptical about
Drawing a plausible schematic is the easy 80%. Hardware punishes the last 20%.
The hard problems in PCB design aren't "what does a Bluetooth controller roughly look like" — a competent model can sketch that. They're signal integrity, EMI, power delivery, thermal behavior, impedance control, and design for manufacturing: the board has to be correct, not just look correct. Then there's component selection in a world where the exact part you specced is out of stock or end-of-life. A generative tool that gets you to a first draft is genuinely useful. A generative tool you can trust to spit out a board that boots, passes EMC, and a contract manufacturer will build without a back-and-forth — that's a much taller order, and it's the one that matters.
So the question for siliXon isn't whether the demo is impressive. It's how much of a real electrical engineer's judgment the system can absorb before the human has to take over — and whether "minutes" survives contact with a board that actually has to ship.
The sovereignty angle
siliXon wraps itself in a now-familiar European story: help the continent rebuild hardware capability and lean less on Asian supply chains. Investors love this framing right now, and System.One leading the round fits the pattern.
It's a good story, and it's worth naming the gap in it. A design tool lowers the barrier to designing boards. It doesn't build fabs, secure components, or fix where things get manufactured — the parts of "hardware sovereignty" that are genuinely hard and capital-intensive. Lowering the design barrier is upstream of all that, and upstream isn't nothing: more European teams able to spin up hardware is a precondition for the rest. But a circuit-board generator is a tool, not a supply chain.
The read
The wedge is real — collapsing PCB design from months to a text box is the kind of barrier-removal that creates new builders, the same way no-code did for the web. The technical bar is brutally high, and the sovereignty narrative is doing some of the fundraising work. $1.5M buys siliXon the runway to find out which of those two facts wins.
The thing to watch isn't the next funding headline. It's the open beta — specifically, how many boards come out the other side that an engineer ships without redrawing.
Automating PCB design through natural language could slash engineering timelines, but commercial viability hinges on proving the AI handles rigorous electrical and manufacturing constraints.
Stance · CautiousConfidence · Emerging
The article recognizes a genuine workflow bottleneck and validates the automation premise, but repeatedly flags unresolved technical rigor and missing adoption data as decisive risks.
Key takeaways
siliXon secured $1.5 million in pre-seed funding to build an AI platform that converts plain-text product descriptions directly into schematics and layouts.
Generating a visual draft is straightforward; the critical hurdle remains guaranteeing signal integrity, electromagnetic compatibility, thermal management, and design-for-manufacturing compliance.
Real-world component shortages and end-of-life parts add complexity that current generative models must navigate to avoid requiring heavy manual intervention.
The European hardware sovereignty narrative drives investor interest but only solves the design friction point, leaving fabrication and supply chain challenges untouched.
