Hello friends, and welcome to The Unsophisticated Investor! Brought to you by Mark, Head of UK Expansion & Operations at Shuttle!

Shuttle builds the regulated rails that syndicate leads, fund managers and their investors use to raise and deploy private capital on. The structuring, the investor checks, the payments, the compliance, the reporting, handled in one place, so the people running rounds can focus on the companies they’re backing, not the admin underneath them.

Now, let’s get into it 👇

Microagi, a Munich robotics company founded just thirteen months ago, says it is now running at $100 million a year in revenue. That would make it the fastest European company to reach the mark. The claim deserves to be taken seriously, but it measures something quite different from the records it says it has broken.

On the 9th of October, Bercan Kilic, Microagi's founder and CEO, posted that the company had gone from 0 to a $100 million annualised run rate in six months, counting from the launch of Shift, its data-collection app. The chart alongside the post placed Microagi ahead of Lovable, OpenAI, Cursor and Wiz. 

Microagi was founded in September 2025. In July it closed a $55 million seed round, reported as the largest in German history. It now claims a level of revenue that took the likes of Synthesia eight years to reach.

In this week's episode we look at what Microagi sells, who is likely to be paying for it, and how much weight the $100 million figure can bear. We also consider where it sits among the European companies that have reached $100 million in ARR the fastest over the last ten years.

Microagi has five co-founders. Kilic was an aerodynamics engineer at Red Bull Racing and Yoan Iliev worked at Mercedes-AMG Petronas F1. Anton Poletaev was a researcher at the Alan Turing Institute, Nico Nussbaum, the CTO, trained as an engineer at RWTH Aachen, and Artjem Weissbeck co-founded the WhatsApp commerce platform Charles.

The Formula 1 background is more relevant than it first appears. A modern F1 team runs on telemetry: hundreds of sensors on the car, with data collected every lap, and engineers turning small observations into small gains. Atlas, Microagi's industrial product, applies much the same discipline to factory robots, collecting data from the production line and using it to tune how a robot behaves at a specific station.

In July the company raised $55 million in seed funding led by Hummingbird, with Northzone, LocalGlobe, Village Global and redalpine also investing. It has opened a research base in Zurich and has teams in Munich, London and New York. 

One funding tracker puts the seed post-money valuation at $275 million. The company has not confirmed that figure, but if it is close, investors in July paid less than three times what Microagi now says is its annual run rate. For comparison, Lovable raised its Series B in December 2025 at $6.6 billion with ARR above $200 million, roughly 30 times revenue, and n8n was valued at about 75 times its $40 million ARR last year. Either Microagi's next round will be priced at a very large step up, or investors will apply a heavy discount to how durable they think this revenue is. 

Microagi is selling into the biggest gap in robotics. Language models were trained on decades of text that people had already put on the internet. There is no equivalent archive for physical work: almost nobody has filmed thousands of hours of towels being folded or dishwashers being loaded from the point of view of the person doing it. Robot models need exactly that footage, and it has to be produced deliberately.

Shift is how Microagi produces it. From late May it offered New Yorkers free two-hour home cleanings, carried out by cleaners wearing head-mounted cameras. The recordings include motion and sensor data and metadata such as GPS location, and that the footage is anonymised before being licensed for AI and robotics training. Shift's launch drew thousands of bookings, and London, Munich and Zurich come next. Plumbing, cooking and building repair are reportedly being considered after that.

The economics run in the opposite direction to a normal cleaning company. Every visit is a cost: the cleaner's pay, travel, supplies, the camera kit and then the work of processing and anonymising the footage. Microagi makes money only if an hour of that footage sells for more than it cost to make. The important detail is that footage, unlike a cleaning, can be sold more than once. Unless its contracts are exclusive, the same hour of video can be licensed to several labs, and the margin on each additional sale is close to total.

The buyers are almost certainly the labs and robotics companies building general-purpose robot models, although Microagi has not named its customers. That market already exists at scale in the United States. micro1, which supplies training data and expert workers to AI labs, has reportedly hit ARR of $700 million. Labs are spending vast sums on data, and Microagi is among the first European companies built to take a share of it.

Atlas, the second product, is a different sort of business. It has been described as a platform that captures a manufacturer's own production data, multiplies it in simulation and fine-tunes robot models for tasks specific to that plant. Microagi does not make robots itself, but works between the robot, the model and the factory floor, which is where most industrial robot projects get stuck. Factory customers are unlikely to let their production data be resold, so Atlas revenue probably looks more like a conventional enterprise contract.

A run rate takes a recent period of revenue and scales it up to a year. A $100 million run rate means Microagi billed something like $8.3 million in its most recent month, or $25 million in its most recent quarter. The company has not publicly confirmed what it's using to calculate  it. Either way, it does not mean $100 million has been received, and it does not mean the revenue will repeat.

For a subscription business like Lovable, that distinction is fairly academic. Revenue comes from a very large number of paying users, and one month is a reasonable guide to the next. A data business sells to a comparatively smaller pool of buyers. A single lab signing a $30 million annual licence would on its own account for almost a third of Microagi's run rate. If some of the revenue comes from one-off purchases of finished datasets rather than ongoing supply, a strong month can be annualised into a figure the following month may not match. Microagi has disclosed neither how much of the run rate is contracted and recurring, nor how it splits between Shift data and Atlas deployments.

The start date also flatters the result. Microagi counts from the launch of Shift rather than from its own founding. Measured from founding it took about thirteen months, which, in all fairness, would still be the fastest on record in Europe. Press coverage dates Shift's New York launch to 28 May, which would make the window closer to four and a half months than six, so the company may be counting from an earlier, unpublicised start. Choosing your own start date is normal. Skild AI, which also appears on Microagi's chart, measured its $100 million from its first commercial deployment.

The figure is not misleading so much as incomplete. What it tells us is that demand for robot training data is large and arriving quickly. It does not yet tell us whether Microagi has built a business that produces $100 million every year, which may come in due course.

Taken at its word, Microagi now leads the European table, whether measured from Shift's launch or from its founding. Lovable has the strongest record that rests on recurring revenue. Each company chooses where its own clock starts, and the revenue behind each $100 million is of a different kind. The type of revenue is critical when it comes to measuring the durability of these revenue figures, and how long it can sustain growth. 

UiPath is the useful reference point from before generative AI. It took a decade to reach its first $1 million in ARR, then grew from $8 million to $150 million between April 2017 and November 2018. At the time that was described as the fastest growth in the history of enterprise software. On today's table it would be mid-table. 

The dividing line is 2023. Synthesia and n8n are both strong companies, but they were founded before generative AI and spent years building towards $100 million. Their growth bent sharply once AI arrived: n8n grew ARR fivefold in the year to March 2025 and went from about $40 million to over €100 million in under ten months. Every company that reached $100 million in under two years was built on the new models from the start.

The fast companies did not all grow the same way, and the differences matter for how long the growth lasts. Lovable and ElevenLabs sold first to individuals and small teams paying by card, then worked their way into larger companies. Revenue from that kind of customer base arrives quickly and is spread thinly, so the loss of any one customer barely registers, but individual users also leave easily if a cheaper or better tool appears.

Legora is the opposite case. It sells to law firms, a slow and conservative customer, and still reached $100 million in under 18 months. The reason is that legal work is priced by the hour, so a tool that saves hours has a price that is easy to justify. Contracts are larger, harder to win and harder to cancel.

Mistral's growth comes from a different source again. Around 60% of its revenue is from Europe, much of it from governments and regulated companies that want a model provider outside the United States. Its customers are buying control over where their data goes as much as model performance.

Microagi does not fit any of these. Its customers are, as far as we can tell, other AI companies, which makes its revenue more concentrated than anyone else on the list and more dependent on how much the large robotics labs choose to spend on data in any given year.

The same few investors appear repeatedly. Hummingbird led Microagi's seed round and was an early investor in Lovable. Accel led Lovable's Series A and Legora's Series D. Most of the value created in European software over the past three years has gone to a small group of funds that were prepared to write large cheques very early.

Finally, how fast a company reaches $100 million says little about what follows. ElevenLabs passed $500 million in ARR by May, according to TechCrunch, and Mistral is aiming for $1 billion this year. For those companies, $100 million was an early point on a much longer curve.

Microagi has done something unusual, but the comparison it has chosen does it few favours. Lovable showed that millions of people would pay every month for a product. Microagi has shown that AI labs will pay a lot for data that cannot be obtained easily elsewhere. That can be a very good business, but it is a more concentrated revenue profile, and investors will likely value it differently from a subscription company with the same headline number.

There are three things worth watching. The first is disclosure: how much of the run rate is under multi-year or recurring contracts, and how much comes from factories through Atlas rather than from footage. The second is the next funding round. At a reported $275 million in July, investors were valuing the company at under three times today's claimed run rate, and the price of the next round will show how much of that revenue the market believes will last.

The third is Europe. Shift's next cities are London, Munich and Zurich. GDPR applies to anyone identifiable on camera, including people who live in the home but never agreed to the recording, and footage inside a home can easily capture sensitive details such as health information or religious belief. If Shift can run here on the same terms as in New York, the data operation can scale across both continents. If it has to be heavily redesigned, the European half of the plan becomes slower and more expensive.

For investors further down the market, the lesson is about when the returns were made. A $55 million seed round is now possible in Germany, and the value in almost every company on this table was secured at seed and Series A, long before anyone was announcing revenue milestones. Getting into those early rounds, and on what terms, is where most of the opportunity sits.

The Unsophisticated Investor is brought to you by Mark, Head of UK Expansion & Operations at Shuttle. Shuttle's mission is to power the next generation of syndicate leads, fund managers, and their investors, backing the great innovations of tomorrow. To do it, we're building private market infrastructure for running investment operations. By stripping away the legal friction and admin for deal managers, we make it easier than ever for Investment managers to run high-performing opportunities and to open those opportunities up to a wider, modern network of investors.

Mark
Shuttle’s Head of UK Expansion & Operations