Decart adds another $32 million to the $500 million-plus price tag

A young startup that emerged from stealth less than two months ago is back in the spotlight, with big-name backers and even bigger ambitions to make a splash in the world of artificial intelligence.

Descartes It is building what CEO and co-founder Dean Leitersdorf describes as a “fully vertically integrated AI research lab” alongside enterprise and consumer products based on the lab’s work. Its first enterprise product that optimizes GPU usage is already generating millions of dollars in revenue. And its first consumer product, a A playable “open world” AI model called OasisIt was released when Decart went undercover and already had “millions” of players.

Now, on the back of a strong out of the gates, Decart has raised another $32 million led by Benchmark.

The Series A funding comes less than two months after the company, which is headquartered in San Francisco but has significant operations in Israel, raised a $21 million investment from Sequoia and Zeev Ventures. in this latest A series.

And TechCrunch understands from sources that Decart’s new post-money valuation is now more than $500 million. (By comparison, seed valued it at just over $100 million.)

Leitersdorf is 26 years young, full of energy and coming fast. The goal, he says, is not just to take on companies we know as big players in AI, like OpenAI, Anthropic, Mistral and others. He said he wanted to build a “kilocorn” — that is, a trillion-dollar company.

“We have a long way to go and we have great things to build on,” he said. However, he noted that yes, the company has already been approached as a takeover target several times. If you look at the optimization piece that Descartes built, there are some interesting (slightly more modest) comparisons, e.g. Run:ai is being acquired by Nvidia for $700 million.

Leitersdorf’s exuberance comes after an impressive ten-year run that got him to where he is now.

Born in Israel, Leitersdorf spent his early years following his parents’ careers (they are doctors and researchers) before moving with his family to Switzerland and then to Palo Alto.

As a junior at Palo Alto High School, he rushed to earn his diploma in just two years, only to return to university, returning to Israel at the Technion, where he completed his bachelor’s, master’s and doctorate in computer science in just one year. five years, including time overlapping with military service.

Its co-founder Moshe Shalev (above, left) is impressive in a different way: he grew up in a strictly Orthodox family and came to computer science while serving time in the IDF. He turned out to be adept at it and helped design, build, and manage AI operations for the IDF’s 8,200 intelligence unit, where he stayed for nearly 14 years.

There is a third co-founder with an equally impressive background, although his name is not being released yet due to current commitments.

Descartes before the horse

Decart, as it exists today, focuses on three main areas, as described by Leitersdorf: systems (currently: infrastructure optimization), models (AI algorithms) and data (you can read these: programs that receive and return data).

Decart’s first product, which it introduced earlier this year while still under wraps, is in the systems camp: software that helps optimize how GPU processes work while training and running inference workloads on AI models.

This software has worked very well: it is used by a number of companies that build and run models to reduce some of the extreme operational costs associated with building or using artificial intelligence. By using the software, workloads that could cost $100 an hour to run can be reduced to just 25 cents an hour, Leitersdorf said.

“It definitely got people’s attention,” he joked. Indeed, AI is very hot right now, but it seems that companies building technology to improve how well AI works are even hotter.

The company won’t name any of its customers, but it claims it’s already generating millions of dollars in revenue and has enough customers using it that Decart turned a profit when it launched in early November. Leitersdorf added that remaining profitable through the end of the year is now on track, and interest from the market is another reason VCs are interested.

“Decart’s innovation makes the generation of artificial intelligence not only more efficient, but also more accessible to any type of user,” said Victor Lazarte, senior partner at Benchmark, who led the deal. “By removing barriers to entry and significantly reducing costs, they are fueling a new wave of creativity and practical applications. We are proud to join them on this journey as they redefine the possibilities of artificial intelligence and its role in our everyday lives.”

It may be the engine that drives the startup so far, but this optimization product is not Decart’s main focus. Leitersdorf said Descartes built it to help finance the business while still in stealth mode. research he did while still a student.

Leitersdorf said Decart’s second product is compatible with the hope of doing more in the future.

Like minecraft Oasisis a “playable” AI that creates AI-based audio and visual interactions that respond in real-time, which it launched to coincide with its out of stealth two months ago.

The plan, Leitersdorf, is to start more experiments along these lines. These include improved Oasis gameplay, generative artificial intelligence and interactivity, among others. These could include AR or VR experiences that don’t require special hardware to run.

“The problem (with VR and AR in the past) was that we started from the hardware rails,” he said. “But it’s hard to build hardware, it’s hard to get people to master new hardware. The great thing about Gen AI is that we can actually (build AR) in the software. We can really bring value before the hardware is ready.”

Ironically, you could argue that one could put the cart before the horse when it comes to some of Descartes’ ambitions. Leitersdorf didn’t have much of an answer to me about how the company’s position would be with respect to customers who want to use optimization software to create or run bad models.

Also, the company does not currently have a plan in place to ensure that the apps it develops are not abused or misused. According to him, at the moment these are not self-presenting scenarios.

The main thing is that more people are interested in his work on the platform, and this activity turns into income.

“The real kingmakers are the users,” Leitersdorf said. “They’re the only ones that matter.”