Physical AI in 2026: Three Signals Worth Tracking
From fresh Series B capital flowing into robot perception software to humanoid robots improvising hugs, the Physical AI market is producing signals worth watching closely.
4 min read
0:00
0:00
From fresh Series B capital flowing into robot perception software to humanoid robots improvising hugs, the Physical AI market is producing signals worth watching closely.
Sereact secured Series B funding to scale Cortex 2.0 beyond bin picking into assembly and kitting, while simultaneously opening a Boston office to enter the U.S. market.
According to The Robot Report, Sereact is using its Series B capital for two parallel moves: expanding the capability scope of its Cortex 2.0 robot brain, and establishing a physical presence in Boston. The expansion from bin picking into assembly and kitting is a meaningful signal. Bin picking is a well-defined, high-volume use case. Assembly and kitting require more adaptive perception, more robust force feedback, and more reliable dexterous manipulation. Investors are apparently betting that Cortex 2.0 can bridge that gap. From a builder perspective, the U.S. market entry is the part worth watching for market structure reasons. Boston sits at the center of U.S. robotics infrastructure, with access to talent pipelines, defense-adjacent contracts, and a concentration of potential enterprise customers.
Bin picking operates in relatively unstructured environments where the robot needs to identify and grasp objects from a pile. Assembly and kitting add constraints: parts must go in specific orientations, into specific slots, in a specific sequence. That requires force control, not just vision. The fact that Sereact is positioning Cortex 2.0 for these tasks suggests the system has moved into territory where perception alone is not sufficient.
Opening a Boston office in 2026 puts Sereact in proximity to a dense cluster of robotics buyers, integrators, and research institutions. The timing aligns with broader capital flows into U.S. physical AI infrastructure. Whether Sereact can compete with established U.S. players on enterprise sales cycles is a separate question, but the geographic bet is defensible on paper.
The Robotics Summit added a dedicated AI-in-robotics track, which is a direct signal that practitioners now treat Physical AI as a distinct discipline requiring focused curriculum.
As reported by The Robot Report, the Robotics Summit and Expo will feature an entire session track dedicated to applying artificial intelligence to robotics. Event programming is a lagging indicator, but it is a useful one. Conferences allocate dedicated tracks when a topic has crossed from emerging curiosity to operational priority. A full AI track at a major robotics summit suggests that enough practitioners are actively deploying or evaluating Physical AI systems that a multi-session curriculum is now justifiable. The practical implication for the audience here: if you are trying to benchmark what the field considers current and relevant, the session list from that track is worth scanning.
A humanoid robot deviated from its programmed dance routine at a Chinese university sports event and initiated a hug with a nearby student, raising immediate questions about behavioral boundaries and system predictability.
Interesting Engineering reported on a scene at a Chinese university sports event where a humanoid robot broke from its scripted choreography and moved to hug a student standing nearby. The incident is easy to frame as a curiosity or a viral moment. From a systems perspective, it is something more specific: a deployed humanoid operating in a live, uncontrolled environment produced an unscripted behavior that involved physical contact with a human. The gap between what the robot was programmed to do and what it actually did is the data point. Whether that gap came from sensor input misinterpretation, a model inference artifact, or something in the interaction design is unclear from the available reporting. What is clear is that the behavior was not planned, and the robot was close enough to a human to make physical contact.
A hug is low-force contact and the outcome here was benign. The concern is not this specific incident but the category it represents. If a humanoid can deviate from scripted motion into unscripted physical contact in a crowd, the question of behavioral envelope definition becomes an engineering priority, not just a policy discussion. Force control, collision detection, and intent modeling all intersect here.
Capital flowing into robot perception software, conference infrastructure formalizing Physical AI as a discipline, and live deployment producing emergent behavior together describe a market that is accelerating faster than its safety and deployment frameworks.
Here is what the data suggests when you look at the three stories together. Investment is moving from narrow, well-scoped robotic tasks toward broader manipulation capabilities, with Sereact's expansion from bin picking to assembly as the clearest example. Simultaneously, the practitioner community is formalizing its knowledge base around Physical AI through structured conference programming. And in the field, deployed humanoids in China are operating in live crowds and producing unscripted behavior. These are not contradictory trends. They describe the same underlying dynamic: the capability frontier is expanding faster than the surrounding infrastructure of testing protocols, behavioral standards, and deployment guidelines. That gap is where the most interesting engineering and policy problems currently sit.
Cortex 2.0 is Sereact's robot brain software platform. According to The Robot Report, the company is expanding its capabilities from bin picking into assembly and kitting tasks, supported by a new Series B funding round and a U.S. market entry via a Boston office.
Physical AI refers to artificial intelligence applied to systems that interact with the physical world, primarily robots. The Robotics Summit dedicated an entire session track to the topic in 2026, as reported by The Robot Report, reflecting that practitioners now treat it as an operational priority rather than a speculative field.
As reported by Interesting Engineering, a humanoid robot at a Chinese university sports event deviated from its programmed dance routine and initiated physical contact with a nearby student. The cause of the behavioral deviation was not confirmed in the available reporting.
Unscripted physical contact in live environments tests the limits of force control, collision detection, and behavioral boundary design. For engineers and investors in the Physical AI space, these incidents are early real-world data on where deployed systems diverge from lab-tested specifications.
Sereact's Boston office opening is one data point suggesting yes. The company is a European robotics software firm using its Series B to establish a U.S. presence, targeting a market where physical AI infrastructure investment is currently concentrated, according to The Robot Report.