Enact Intelligence Raises $30M: Every Robot Motor Becomes Force Sensor

Automatically extracted; no manual editorial approval recorded.

e. LeCun's investment is not incidental to his research convictions. His entire scientific thesis — the argument he has been making since leaving Meta at the end of 2025 and then raising $1.03 billion for AMI Labs in March 2026 — is that AI systems need to learn about the physical world through sensorimotor experience, not language alone. A robotics company that generates proprioceptive force data as its primary trai

Open source S12

ngly span both worlds. Kamil Mieczakowski of Notion Capital framed the rationale directly: robots will not generalize until they have a nervous system, one that lets them not only see the world but feel it. redalpine general partner Harald Nieder described the investment as a data-layer bet: Enact is tackling dexterity with a new data modality , starting with high-precision force data from bespoke actuators. What Mak

Open source S14

es This Architecture Different From Its Closest Competitors The robotics manipulation field has two other well-funded approaches to force sensing that reached major milestones the same week Enact went public. Eka Robotics, a Cambridge, Massachusetts startup founded by MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja, raised $150 million at a $1.45 billion valuation on October 20, led

Open source S15

gertip resolution. University of Illinois robotics professor Kris Hauser raised a question that applies equally to all three approaches when assessing Eka: whether the technique works reliably on real assembly lines for a year, and whether it generalizes beyond picking up and placing objects. Enact has not yet published success rates, cycle times, or comparative benchmarks for any of its claimed tasks. The Motor-Curr

Open source S16

uperposition of contact-dependent signal and internal noise: gearbox friction, inertia, temperature drift, Coulomb friction at low speed, backlash, and cable tension all contribute to current draw independently of external contact. In standard off-the-shelf robots with high gear-reduction ratios optimized for position accuracy, these internal contributions can bury the contact signal entirely. This is why Enact is de

Open source S17

signing its own actuators rather than buying them commercially. By minimizing internal friction and backlash in the actuator design — a family of approaches known as backdrivable or quasi-direct-drive actuation, used in research platforms like MIT's Cheetah legged robot — the company can improve the signal-to-noise ratio of the contact signal in the motor current. Job postings reviewed by Nexithon Technologies' analy

Open source S18

sis suggest the team is also developing FPGA-based control electronics , which would give them the low-latency signal processing needed to separate contact-related current changes from internal dynamics in real time. The constraint this creates is significant: Enact cannot easily decouple its sensing capability from its hardware. A model trained on one robot's actuator-current signature does not automatically transfe

Open source S19

r to a different arm with different gearing and friction characteristics. The company says its approach is form-factor agnostic — that the same intelligence applies across robot shapes — but this claim depends on the learned models generalizing across hardware variants, which has not yet been demonstrated publicly. The Zurich Bet and the SME Gap Galliker made a deliberate decision to build in Zurich rather than Silic

Open source S20