Frenetic

A small room in Madrid that is quietly rewiring the world

It is a Tuesday morning on Avenida de Córdoba, and somewhere on the third floor a power-electronics engineer in Stuttgart is uploading a spec sheet to a browser tab. By the time her coffee cools, a cloud platform on the other side of the continent has proposed a transformer, run a quasi-3D simulation of its losses, picked a core material, suggested a winding, and quoted her a physical sample. Two weeks ago this work would have eaten a quarter of her year. Today it is a side errand.

This is what Frenetic has done to magnetics, and magnetics, it turns out, is what holds the energy transition together.

A 1970s craft hiding inside 2026 hardware

Transformers and inductors do not sound like a venture-scale problem. They are not. They are a much harder one. Power electronics is the layer of the world that converts one kind of electricity into another - high voltage to low, AC to DC, battery to motor - and inside every converter there is a magnetic component custom-built to the application. There is no Amazon for these things. There is, mostly, a spreadsheet, a senior engineer who has been doing it since before broadband, and a queue of prototypes that take months to land on a desk.

The conventional process, per Frenetic's own pitch, can run up to nine months. Nine months is a long time when the world has decided to electrify everything by Tuesday. The auto industry wants smaller, cooler, denser magnetics for EVs. Aerospace wants them lighter. Data centers want them by yesterday. The senior engineer cannot scale.

Two engineers from Politécnica de Madrid who refused the spreadsheet

Frenetic was founded in 2015 by Chema Molina, a PhD in industrial electronics out of Universidad Politécnica de Madrid - one of those quiet European departments that has been turning out magnetics specialists for decades while no one was looking. He was joined by Alfonso Martínez de la Torre, another Politécnica engineer with practical years inside the magnetics industry.

Their bet was simple and slightly absurd. Build a database large enough, calibrate it against real lab measurements, and let machine learning do what the senior engineer's intuition has been doing for forty years - only faster, with receipts, and without taking holidays. The investors who eventually wrote the cheques did so because the bet, if it worked, would touch nearly every wire that carries serious current.

The cast

Group portrait, if anyone could ever convince them to sit for one.

Three tools and a small factory

The platform splits cleanly into four things. There is the Frenetic Simulator, which takes a magnetic component, runs loss models, leakage inductance calculations, and thermal analysis, and returns answers in the time it takes most simulators to load. There is Frenetic AI, an early-stage assistant that suggests power supply topologies and recommends magnetics before a sketch is even on the back of an envelope. There is Frenetic Planar, dedicated to planar transformers and the finite-element analysis they require. And there is Frenetic Factory, opened in 2024, which translates a finished design into a physical sample on a bench in days.

The interesting part is not any one of these. It is that they share a calibrated database. Most magnetics tools live or die by the accuracy of their core-loss equations. Frenetic's models are trained on, and tested against, actual lab measurements - which is why they will tell you about 3x improvements over the analytical methods that have been quoted in datasheets since the Carter administration.

Logos arranged in a row, as is customary. Notice how the row contains both a space agency and a kitchen-appliance giant.

The numbers behind the pitch

Pitches make claims. Engineers want bars on a chart. Frenetic's deck has both. What follows is, by the company's own published numbers, what their platform does to the average magnetic design cycle.

It is worth noting that none of this is magic. The reason it works is that magnetics, despite its mystique, is a regime where data plus careful measurement beats analytical shortcuts almost every time. Frenetic's edge is the database and the discipline. The AI is the wrapper.

"Give engineers magnetic design powers"

Frenetic's own tagline reads "Give Engineers Magnetic Design Powers - and Decision Makers the ROI They Dream of." The first half is the part the engineers respond to. The second half is the part that closes contracts.

Underneath the marketing, the mission is narrower and more interesting. Magnetics is one of the few remaining engineering disciplines where a small team of specialists gates the speed of an entire industry. Frenetic is trying to remove that gate without removing the specialists - by giving them tools that let one person do what used to take five, and by giving the people who used to wait for them a way to skip the queue.

The company is B2B, not consumer. It sells subscriptions to power-electronics teams at OEMs, plus a manufacturing service for prototypes and short production runs. There is no influencer push. There is, mostly, a YouTube channel with painfully practical product walkthroughs and a LinkedIn page that posts technical notes.

Why this matters more in 2030 than in 2026

The case for Frenetic is not really about software. It is about the physics of the next decade. Every projection for electrification - EVs, data centers, heat pumps, grid storage, electric aircraft - depends on cramming more power through smaller, hotter, faster components. Magnetic design becomes the chokepoint of the chokepoint. The companies that can iterate magnetics in weeks instead of months will ship products in years instead of decades.

Frenetic is small. Fifty-four people. Three million dollars or so in revenue, by current estimates. It is not the kind of company that shows up on consumer headlines. It is the kind of company that, ten years from now, will quietly appear in the supply chain of nearly every electrified product you own - which, ten years from now, is going to be most of them.

Back to the Tuesday morning

Return to that engineer in Stuttgart, the one whose coffee is still warm. Three years ago she would have written an email to a magnetics specialist, attached a spec sheet, and gone on holiday. The specialist would have come back with options. They would have ordered prototypes. The prototypes would have come back wrong, or right enough, or right but late. A senior engineer somewhere would have done the rest by hand.

Today she has a simulator, a topology suggestion, and a sample arriving on Friday. The senior engineer is still there - Frenetic insists this matters - but the queue is gone. Madrid The room in Madrid is still small. The world outside it is moving faster.