Station F, a Paris-based startup hub founded by French billionaire Xavier Niel, is gearing up for a new edition of its F/ai accelerator program in a bid to strengthen its positioning as a stepping stone for promising AI startups.

Launched in January of this year, F/ai’s planning to kick-start its second batch this September, aiming to help a handful of AI-focused startups move from early product to real revenue in a matter of weeks.

Spanning 538,000 square feet, Station F is often described as a co-working space, but its footprint extends beyond the physical space, its director Roxanne Varza told TechCrunch.

One example is Station F’s Future 40 annual selection, in which the team names the most promising teams among some 1,000 companies it welcomes each year. In 2024, TechCrunch observed that nearly all of that annual cohort incorporated AI into its core business.

Station F today has a front row seat to the rise of AI startups, leveraging its position as a cornerstone of “la French Tech.” The startup hub has also successfully leveraged its position to capture equity stakes in its Future 40 companies. “We have been investing [in these companies] since 2022,” Varza said.

Helped both by its size and Niel’s connections, Station F has become a frequent stop for officials seeking to connect with Europe’s tech scene, with no less than 11 presidential visits since President Macron’s inaugural tour in 2017. It has also welcomed AI big names like Sam Altman, and is now leveraging these ties for F/ai.

The first cohort of F/ai’s program was backed by a long list of significant tech companies — AMD, Anthropic, AWS, Clay, Google, G42, Hugging Face, Lovable, Meta, Microsoft, Mistral AI, OpenAI, OVHcloud, Snowflake, and Qualcomm — not to mention several VC funds.

The second cohort will add a few more big names, TechCrunch has learned: Eleven Labs, Nebius, Rippling, OpenRouter, Hubspot, and Github.

“The goal was to bring together all the major players and make it much easier for [AI] startups looking to launch in Europe to connect with them,” Varza said.

Two teams from the accelerator’s first batch have already gained international recognition: Alpic, which won the global grand finale of The Pitch, a competition organized by Deel; and Rippletide, which won the OpenAI Codex Hackathon.

While awards rarely hurt, especially when they bring funding, F/ai is focused on helping its cohort generate revenue, targeting €1 million (about $1.14 million) within six months. “We’d heard quite a bit of criticism about the slow pace of commercialization of European startups,” Varza said. “This brings them on par with what investors are seeing in the U.S.”

Investors seem to like what they’ve seen so far. The first cohort collectively raised $34 million in pre-seed funding, according to Station F. The teams’ track record may have also helped: 80% of these 20 AI startups were founded by repeat entrepreneurs, a third of whom hold PhDs.

The founder profile skews that way mostly because F/ai selects its cohort exclusively via recommendations from founders, partners and investors — a process that could add to the cliquishness and elitism France’s tech scene is at times accused of.

But while teams can’t apply directly, they can get in touch with one of F/ai’s many partners, and perhaps soon with alumni, Varza said. She added that Station F has some 30 other programs startups can apply to.

Access appears to be a key focus for F/ai, which has in the past hosted the likes of Turing Award winner Yann LeCun for private chats. “Today, if the founders here want to speak to people at this level, they all seem to think they need to go to the U.S. and join a program there. We actually want to show that you can stay here and do it from here,” Varza said.

Meta and Snap rolled out new smart glasses last month, the latest sign that the industry is racing to put a camera and an AI assistant onto users’ faces. As the fast-growing market heats up, upstarts like Even Realities are muscling in on the giants.

Even Realities, a three-year-old Shenzhen-headquartered startup, has raised $150 million in a pre-Series B round led by Meituan and previous backer Tencent; the round valued the startup at $1 billion valuation. Founder and CEO Will Wang told TechCrunch that while rivals chase camera-equipped devices built around content capture and AI, his company is betting on display-first glasses that beam information straight into the wearer’s line of sight without giving up privacy.

Even’s earlier backers are mostly high-profile China names — Hillhouse, Sequoia China, and Northern Light Venture Capital.

Even was started by ex-Apple engineers in 2023. CEO Wang worked on the Apple Watch and iPhone; other co-founders came from tech, and two came from luxury eyewear companies, including Lindberg. The startup moved quickly, launching its first product, the G1, in 2024 as what Wang calls the lightest waveguide smart glasses then on the market.

Even blew past its own 10,000-unit target to become the first company in the category to sell more than 10,000 pairs, according to the company CEO. It raised money faster than expected, and swelled from 30–40 staff in 2024 to 300–400 today.

The startup’s latest flagship, the G2, hit the market last November and skips the camera entirely. Instead, a heads-up display built into the frames feeds information to the wearer, controlled by a companion ring, the Even R1, that users tap and swipe to navigate.

Removing the camera is an important part of Even’s privacy philosophy, though not the entire story, Wang continued. Smart glasses, he said, are probably the most personal computing device people will ever wear. Worn on the face all day, they have to feel comfortable to both the wearer and those around them, so privacy is designed into both the hardware and the software. Voice features like translation transcribe audio into text rather than storing recordings; user data is encrypted, and the infrastructure is built to meet Europe’s strict privacy standards, Wang added.

Even’s power users lean hard on Conversate, a copilot that reads a conversation in real time, explaining unfamiliar jargon or feeding follow-ups on the fly, then syncing a summary to their phone.

Still, Even has invested most heavily in optics (the display and overall optical performance), which Wang says is what separates smart glasses from other consumer electronics.

“With a phone or a watch, the display is just a conventional OLED or LCD screen. Smart glasses are the first product category to rely on optical displays, which require an entirely different technology stack; you have to design the microchip, the optics, and the waveguide together. That’s where we’ve invested the most,” Wang said.

The company developed a proprietary optical technology called Even HAO, or Holistic Adaptive Optics, an end-to-end design that integrates the microchip, waveguide and prescription support from the start, rather than combining components designed separately.  

More than half of Even’s users sit in the U.S. — its fastest-growing market — and so does the bulk of its developer community. The company doesn’t sell in China yet, even though it manufactures there across several factories; its main markets are the U.S., Japan, South Korea, the Middle East, and Europe. “The demand there is significant, so we want to make sure we’re prepared first,” Wang said.

Even sells near the top of the category on price and still moves real volume, making it a profitable player in the space, Wang said. “Most of our customers are male professionals between 30 and 50 years old. We ran a survey and found that about a third of our users are company executives,” he added. The frames retail for $599 before tax; prescription lenses or the ring tack on another $200–$300, pushing the average order to roughly $1,000.

The humanoid robotics market is awash in money right now. Last week, AI2 Robotics, a Shenzhen-based startup that makes wheeled humanoid robots, raised nearly $735 million at a nearly $3 billion valuation. Earlier this year, Apptronik, an Austin-based maker of humanoid robots for manufacturing and logistics, closed a $935 million funding round valuing the company at more than $5.5 billion, backed by Google, Mercedes-Benz, and John Deere, among others. Last fall, Figure AI, a San Jose-based startup developing general-purpose humanoid robots, self-reported that it closed on $1 billion in Series C funding at an eye-popping $39 billion valuation.

By comparison, Peggy Johnson, CEO of Agility Robotics, is surprisingly measured. We spoke by phone last week, just after the company announced plans to go public through a merger with Michael Klein’s Churchill Capital Corp XI, a special purpose acquisition company, or SPAC. The deal values Agility at around $2.5 billion and is expected to raise more than $620 million in gross proceeds, the largest capital raise in humanoid robotics history. It hasn’t closed yet; the merger still needs shareholder approval and SEC review, and is expected to be completed later this year.

Agility was founded in 2015 as a spinoff from Oregon State University. Based in Salem, Oregon, the company makes bipedal humanoid robots designed to work in warehouses and factories. Its move is notable for a few reasons. It would make Agility the first pure-play humanoid robotics company to trade on public markets, giving retail investors direct exposure to a sector that has so far been available primarily to deep-pocketed VC funds. It also offers a rare window into the finances of a business in a space where most competitors closely guard their numbers and even the state of the tech they are building.

Johnson — formerly executive vice president of business development at Microsoft, where she helped engineer the $26 billion acquisition of LinkedIn, and later CEO of Magic Leap, the once-hyped augmented reality headset maker — was careful throughout our conversation. She declined to offer forward-looking financial guidance, declined to disclose the bill of materials for Agility’s flagship robot Digit, and pushed back politely whenever questions veered toward speculation.

Asked why Agility is going public via a SPAC rather than raising another private round — a structure that skips the roadshow and pricing scrutiny of a traditional IPO — Johnson said much of it boils down to the first-mover advantage the company enjoys when it’s the first of its ilk to go public. For investors clamoring for shares in a buzzy robotics company, Agility is “an acceleration story and a timing story,” she said. The proceeds will also help Agility ramp up production at its 70,000-square-foot manufacturing facility in Salem, Oregon, and fulfill an existing pipeline of customer orders.

As for the troubled reputation of SPACs — many companies that went public that way in 2021 famously fizzled out entirely or trade well below their offering price — Johnson was unfazed. “If we just keep our head down, keep delivering customer by customer, robot by robot, we hopefully won’t experience the same volatility,” she said. “Our biggest competitor right now is just us. How quickly we can execute, how quickly we can continue to add new skills.”

The pipeline goes well beyond pilots, Johnson told TechCrunch, pointing to more than $300 million in booked, multi-year revenue that represents roughly 1,000 robots that are part of a robots-as-a-service model in which customers pay a monthly fee rather than purchasing the machines outright. “Everybody on our list right now is already vetted, and they have deployment plans behind their proof of concepts,” Johnson said. Customers include GXO Logistics, Amazon, Toyota Motor Manufacturing Canada, Schaeffler, and Mercado Libre.

Digit itself is a deliberately unfussy piece of hardware. It stands about 5’9″, weighs around 160 pounds, and is designed to do one thing exceptionally well, which is move heavy objects in human-built spaces. Its most distinctive feature is a set of reverse-bend knees — they’ve been called “bird legs” — that allow it to reach from floor level to overhead shelving without the knees colliding with warehouse racking. (Agility’s founders, Johnson explained, weren’t interested in biomimicry for its own sake.) The robot’s hands — two thumbs and two fingers — are similarly task-specific; they’re optimized for gripping heavy plastic totes, even as their contents shift in transit.

Johnson said Agility is “LLM-agnostic,” drawing on models including Claude and Gemini to handle what she calls the semantic layer — translating high-level instructions into robot behavior. She described a recent test in which engineers scattered different types of trash on the floor and told Digit simply to “clean up this mess.” The robot assessed, sorted, and binned everything correctly, including correctly identifying bubble wrap as non-recyclable.

Of course, it’s the physical layer — the mechanics of balance, locomotion, and manipulation — that Agility considers its core proprietary advantage, one built up over more than a decade of real-world deployment. “The LLMs had the entire internet to train on,” she said. “When you think about the physical AI of humanoids — that doesn’t quite exist yet.” At most companies, anyway. Johnson believes Agility is the exception: “We may have the largest data lake of actual operating robotics data in real-world environments.”

Beyond raw data, Johnson said, safety is where the gulf between Agility and its competitors is biggest and most consequential. While rival companies showcase their robots in lab demos and choreographed videos, Agility has had to meet actual industrial safety certification requirements to operate inside customer facilities. “You can’t build your robot and then make it safe,” she said. “That’s a redesign. You have to have all of the safety certified — the electrical system, all of the parts, and the software to support all of that.” (It’s not a trivial concern given that humans are often somewhere in the room. Back in November, Figure AI’s former head of product safety sued the company, alleging he was fired after raising concerns that its robots were powerful enough to fracture a human skull. Figure has disputed the claims.)

As for the home, Johnson thinks humanoids will get there eventually, but she said not to expect them to deliver breakfast in bed anytime soon. It’ll be “10-plus years,” she said of the timeline, observing that warehouses and factories, for all their complexity, have fixed aisles and predictable equipment and workflows unlike homes that are chaotic, with dogs, babies, visitors, and objects left in unexpected places.

“At least roads have some discipline to them,” Johnson added, comparing the challenge to that of autonomous vehicles. “Most of the areas that humanoids will be operating in don’t.”

Agility isn’t ruling out the home market. Johnson said the company will enter it when it makes sense. For now, though, it’s laser focused on the warehouse market, given the growing numbers of retiring workers and younger workers who aren’t willing to take physically demanding roles. “There’s something like over a million jobs in the US today in these areas that are unfilled,” she said. “They’re just very, very hard to hire for.”