While OpenAI pursues yet another monster fundraise, it is not stealing all the oxygen in the room: AI startups building promising foundational models can still open doors, and checkbooks. Multiple sources tell us that Black Forest Labs — a startup that’s building generative AI image models and came out of stealth two months ago with $31 million in funding — is closing new funding. A $100 million round at a $1 billion valuation is the amount we are hearing. The deal may not be final and so could still be subject to change.
Black Forest is not just any AI startup: The company was co-founded by the engineers who built the technology behind Stability AI. And it has a big-name customer. Elon Musk’s X.ai is using Black Forest’s Flux.1 text-to-image model to power image generation in its Grok chatbot. That’s a service that set people chattering immediately after it was launched in part because of the audacious results people generated with it.
“No filters” still appears to be a thing a month later. We created the image on the right earlier this week.
The company is also catching the eye of investors because of its founders and founding team. They include Andreas Blattmann, Patrick Esser, Dominik Lorenz and CEO Robin Rombach, researchers who created Stability AI, considered a game-changing platform for image generation.
“Robin Rombach is known to be an absolute expert at image diffusion models and when you have someone that smart and proven in a brand new space, it makes it obvious one should invest if given the chance,” one of the company’s investors told TechCrunch.
It’s not completely clear yet who is investing in the Freiburg, Germany-based startup’s latest round. One source mentioned that Lightspeed — one of the more prolific investors in AI in Europe, backing Helsing, Mistral, Stability AI and others — might be involved. Lightspeed has not yet responded to a request for comment, and neither has Black Forest itself. (We will update the post if they do.)
The company’s previous, $31 million round included a high-wattage list of investors. Led by Andreessen Horowitz, others, per PitchBook data, included General Catalyst and Stuttgart VC Mätch.vc, with Nvidia’s Timo Aila, Oculus co-founder Brendan Iribe, Apple AI research scientist Vladlen Koltun, entertainment mogul Michael Ovitz, and Y Combinator’s Garry Tan also in the mix.
The $1 billion valuation is a big jump on its post-money valuation from that last round, which was a more modest $150 million. (Asked about more funding, Andreessen Horowitz declined to comment for this story.)
Rapid fundraising in the area of generative AI has become quite commonplace in the current market: Startups building these tools need the funding to buy compute, to hire talent, perhaps to settle IP licensing agreements, and for marketing and business development to compete against bigger and even more well-funded players. In the case of Black Forest Labs, there are more technology launches coming up soon. The company has already said it’s working on a state-of-the-art text-to-video tool, with an as-yet-unannounced debut date.
But the market has been tricky and sometimes unkind to some of the smaller AI players that have raised a lot and now have pressure to deliver. H in Paris, a generative AI startup started by DeepMind alums, raised $220 million in May of this year. It has already lost three of its five co-founders, allegedly over operational differences. Aleph Alpha, which has raised more than $500 million, appears to have pivoted to enterprise services over building foundational models.
“Getting into a headline position, being put into the spotlight, but not delivering,” was how another investor who talked to TechCrunch described the predicament that companies like Aleph Alpha and H have faced.
Black Forest Labs will naturally try to avoid such issues, especially since — at least for the moment — it lacks a strategic investor that might prop it up with giant amounts of cash to grow more aggressively. Said this same investor: “I think they will try to go down the other road, the one of staying as secretive as possible.”
Source : Techcrunch
