Luffy AI Raises $10.4M Series A for Neuroplastic AI Control

Jan de Vries · 2026-07-07

A UK startup just landed a big funding round to bring real-time AI smarts into the physical world—think motors, pumps, and robots that can learn and adapt on the fly. Here's the full story. ### The Funding Round Luffy AI, based in Oxfordshire, England, has raised $10.4 million (converted from €9.4 million) in a Series A round. The company focuses on neuroplastic AI for real-time adaptive control, and it plans to use this cash to speed up its commercial rollout. The round was led by BGF, with participation from MIG Capital AG through its MIG Fonds. Existing investors Bow Capital, Chrysalix, Momenta, and UKI2S also chipped in. ### Why This Matters Dr. Matthew Carr, co-founder and CEO of Luffy AI, put it simply: "AI has been transformative for language and image generation, but has yet to make a substantial impact in industry beyond predictive maintenance and dashboards." He added that factories, motors, and physical systems need AI that's small, fast, and adaptive in real time—not cloud-dependent or hungry for huge data and compute resources. "At Luffy, we've already proven what's possible with AI motor control," Carr said. "We'll use this new funding to scale up our delivery and rollout." ### The Technology Behind It Founded in 2019 by Dr. Carr and Dr. Alex Meakins, Luffy AI is building what it calls the control layer for physical AI. The company argues that industrial AI adoption is held back by the data, compute, and connectivity demands of conventional deep learning. Their solution? A neuroplastic AI stack that excels in real-time adaptive control. Here's what makes it different: - Sparse neural networks trained in simulation, without large datasets - Refined in reality, achieving up to 400x greater efficiency than traditional deep learning - Lightweight, energy-efficient, and self-improving—no constant retraining from the cloud needed The company's Adaptive Neural Controllers (ANCs) learn the physics of a system from first principles and adapt autonomously in real time. They run on constrained hardware already in millions of devices, at timescales that make conventional deep learning look "Big & Slow." ### Real-World Performance Luffy AI's ANCs have been benchmarked against the Google DeepMind Real World RL Suite. The results? 800x fewer synapses and 400x less compute required for equivalent or better performance on tasks. That's a massive leap in efficiency. The company says its models are ideal for complex edge use cases, like industrial motors, VFDs (variable frequency drives), thermal control, and robotics. Right now, they're deploying AI into industrial motor control and VFD applications—think pumps, fans, and conveyors. ### A Big Opportunity Here's a staggering stat: about 50% of the world's electrical energy is consumed by electric motors, and the vast majority of those motors are inefficient. Luffy's adaptive AI-based motor control could enable plug-and-play motors that tune themselves to the load and operating conditions. That means energy savings, reduced commissioning time, and better overall performance. ### What Investors Say Kate Ronayne, early-stage investor at BGF, summed it up: "Luffy AI is disrupting an industry norm that has stood for 100 years. Embedding highly specialized AI directly into physical industrial systems reduces reliance on specialist engineers through a self-commissioning, one-size-fits-all approach." She added that the company has taken impressive steps to validate its technology, and BGF is delighted to partner with them as they scale. ### What's Next Luffy plans to use the fresh capital to drive its commercialization pipeline, moving successful proof-of-concepts and pilots into significant partnerships with leading industry brands. In the longer term, the technology could support a wide range of use cases, including positioning control for robotics and drones, thermal process control, and physical AI applications. The future of industrial AI is looking a lot more adaptive—and a lot less cloud-dependent.