Nvidia’s days of unparalleled market dominance aren’t over, but challengers and options are emerging from all directions.
ZMLa hot French AI startup was approved by Turing Award winner Yann LeCun released inference rendering software that allows a variety of open source models of major languages; to run to a variety of crisps — including Nvidia, AMD, Google’s TPU, Apple Metal and Intel Arc.
With ZML/LLMDthe recently launched LLM inference server, the company’s ambition is to break down existing silos and make different chips available for AI use cases at their maximum available speed and sometimes faster, ZML founder Steeve Morin told TechCrunch.
As AI integrates into work and our daily lives, inference optimization — also known as prompt processing — has overtaken model training in importance, but often feels piecemeal behind the scenes, with software and architecture barriers leading to vendor lock-in, Morin said.
The promise of peak performance on a variety of chips is a technological achievement, but it could also disrupt the market amid growing fears about the costs associated with artificial intelligence.
ZML hopes to give enterprises and clouds the option to use a mix of chips, some of which may be less expensive or use less power. “The idea is to give people back the power to create their own system and achieve real efficiency gains that enable [AI] to spread,” Morin said.
Such software assistance can help new AI chip makers, many of whom happen to be from Europe, Morin noted, citing Axelera, Fragility, Kalray, OLIX, Q.ANT, SiPearl, SpiNNcloudand VSORA. But more than their home region, what matters to him is that ZML can work with them on “things that have never been done anywhere else in the world.”
That’s not to say Morin is down on Nvidia. Is notpartly because of its existing supply. He told TechCrunch that ZML has a good relationship with the AI chip giant, which is preparing for the rise of inferences.
The bottom line was an area of such intense investment that the trend was hailed the “gold rush conclusion.” So ZML has competition like Baseten, recently valued at $13 billion. In conclusion, from the creators of the open source project vLLM; as well as RadixArkthe trading company behind SGLang.
Both vLLM and SGLang partially compete with LLMD, but Morin’s ambitions for ZML cover a wider spectrum. “We’ve gotten to the point where we’re co-designing silicon,” he said. He further credited ZML’s lean team of 20 as the reason the Paris-based startup has been able to move quickly, with more releases in the plans.
It also helped that this small group is well funded for its size. Thanks to his background as VP of engineering at Zenly, which acquired Snapchat for nine figures in 2017, Morin has raised $20 million from venture firms including Harry Stebbings’ 20VC, >commit, AALVC, Drysdale Ventures, Xavier Niel’s Kima Ventures, Puzzle, and Londrcal.
Unlike ZML’s first public project, the conclusion was focused ML framework released in 2024 and updated in MarchZML/LLMD is not open source. But it is released as a free product with the aim of learning to use. “I prefer to count and [then generate revenue] where it’s more efficient without stunting my growth stupidly because I was too greedy to begin with,” Morin said.
It is too early to say when ZML/LLMD might become a paid product and what its adoption will be like. But the startup’s capital board confirms that other founders are paying attention, including Dagger and Docker founder Solomon Hykes, Hugging Face’s Clément Delangue and Julien Chaumond, as well as LeCun, now with AMI Labs. This also reinforces the premise that European AI startups can now build from home. “I couldn’t do ZML anywhere but Paris,” said Morin.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
