- Paris-based NP Company has raised €6M in pre-seed funding led by Partech, with backing from the co-founders of Mistral AI, Dataiku, and Artefact.
- The Inria-born startup is building AI-native physics simulators that deliver results in seconds rather than days targeting a global engineering simulation market projected to reach $30B by 2030.
- The raise is notable for who’s in it: Mistral AI’s own co-founders backed NP Co. just weeks after Mistral acquired Emmi AI, a direct competitor in the same space.
On May 19, Mistral AI announced it was acquiring Emmi AI, an Austrian startup building physics AI models for industrial simulation across aerospace, automotive, and semiconductors. Thirteen days later, Mistral co-founders wrote a cheque into a Paris startup doing almost exactly the same thing.
NP Company has raised €6 million in pre-seed funding to build a new class of physics simulator powered by transformer models pre-trained on industrial physics. The round was led byPartech, with participation from the Peugeot family office and angel investors including Mistral AI co-founders Guillaume Lample and Cédric O, Dataiku founder Florian Douetteau, and Artefact founder and CEO Vincent Luciani.
From Inria research to industrial AI
NP Company was founded in 2025 by Emmanuel Menier and Matthieu Nastorg, both PhD researchers in AI for simulation at Paris-Saclay University, out of Inria – the same French research institute that incubated Mistral AI itself. The company is headquartered in Paris and is currently hiring AI researchers and engineers.
Where legacy simulators solve every engineering problem from scratch – a process that can consume days per design iteration – NP Co.’s models arrive pre-trained on industrial physics, generating high-fidelity results in seconds. The company has already demonstrated 1,000x speedups on industrial benchmarks organised by aerospace giant Safran, with a stated path to 50,000x on full-assembly simulations.
“Engineering is the next frontier for AI, not the next chatbot,” said Menier. “We’re removing that bottleneck, so engineers can spend their time solving humanity’s most urgent challenges, not waiting for a simulation to resolve.”
The contradiction at the heart of this round
Mistral’s acquisition of Emmi AI was framed as a move to build the leading AI stack for industrial engineering, combining Mistral’s language model platform with Emmi’s physics simulation expertise. More than 30 of Emmi’s researchers and engineers joined Mistral’s Science and Applied AI teams in May 2026.
So why are two of Mistral’s co-founders now backing NP Co., which operates in the same space? Three readings are possible. The first is pure conviction: Lample and Cédric O may believe the AI simulation market is large enough to support multiple approaches, and that NP Co.’s specific technical architecture, transformer models pre-trained on physics rather than adapted post-hoc is meaningfully different from what Mistral-Emmi is building. The second is a hedge: personal angel investments are separate from corporate strategy, and backing a competitor’s competitor is a rational way to stay close to a fast-moving technical frontier. The third, and most interesting, is that the Inria connection matters, NP Co. emerged from the same research environment that produced Mistral, and Lample and Cédric O may simply trust the founding team more than the headline suggests.
Menier offered a straightforward explanation. “Emmi was exploring options, and we were not on the market, having just come out of the round,” he told TFN. The distinction matters: Mistral’s acquisition of Emmi may have been less a strategic choice between two rivals and more a function of which company was available.
The investor backing it
Boris Golden, Partner at Partech, which manages €2.5 billion across 220 companies including unicorns Xendit, Alan, and Sorare led the round. Partech has backed NP Co. from inception.
The competitive landscape
NP Company enters a field where capital is already concentrating fast. As TFN previously covered, London-based PhysicsX, founded by ex-Formula One engineers from Renault and Mercedes has raised $155 million in total, including a $135M Series B led by Atomico in June 2025 and a $20M extension from Nvidia, valuing it at close to $1 billion. Rescale secured $115 million in a Series D in April 2025 to modernise high-performance computing and simulation infrastructure. Where both operate largely at the infrastructure or workflow layer, NP Co. is rebuilding the simulation engine itself using foundation models trained on physics, a fundamentally different technical bet, and the same one Mistral just paid an undisclosed sum to acquire through Emmi.
According to MarketsandMarkets, the global simulation software market stood at $19.95 billion in 2024 and is projected to reach $36.22 billion by 2030, growing at a CAGR of 10.4%.
The deeper question this round surfaces is not whether NP Co. can compete, it is whether Mistral’s co-founders see something in NP Co.’s architecture that Emmi, now absorbed into a larger platform, can no longer pursue independently. If so, the most interesting AI simulation company in Europe right now may not be the one Mistral bought.