Physical AI, the technology that combines artificial intelligence with robots that interact with the real world, is moving from experimental prototypes into actual production and classrooms. Three major developments this week show the field accelerating toward mainstream adoption: Chinese automaker Xpeng is ramping up humanoid robot manufacturing, US schools are launching a nationwide robotics education initiative, and a startup is operating a 24/7 interactive robot accessible to the global public.

Why Are Automakers Suddenly Building Humanoid Robots?

Xpeng's humanoid robot, called IRON, has entered small-batch trial production at its Guangzhou factory, with the mass production line now in final testing phases. This marks a significant milestone because automakers like Xpeng possess a critical advantage: they have spent years developing the perception systems, motion control algorithms, and manufacturing infrastructure needed for autonomous vehicles. Those same capabilities transfer directly to humanoid robotics.

The logic is straightforward. Smart cars constantly perceive their environment, understand complex scenarios, and make real-time decisions. Humanoid robots face similar challenges, but in more unpredictable human and industrial environments. Xpeng's cameras, LiDAR sensors, AI algorithms, motion control systems, and electric drive technology all apply to both vehicles and robots. Tesla has already deployed this strategy, with 50 units of its Optimus Gen-3 robot now working at Tesla's Shanghai Gigafactory handling seat installation, interior assembly, component handling, and quality inspection.

For Xpeng, the near-term applications are clear: factories and retail environments. The company has partnered with Baosteel, a major steel manufacturer, to deploy IRON robots for inspection and other complex industrial tasks. Xpeng also plans to deploy IRON as shopping guides in its retail stores starting in the first quarter of 2027. These real-world deployments serve a dual purpose: they reduce human workload while generating the massive amounts of data needed to train and improve the robots' intelligence over time.

How Are Schools Preparing the Next Generation for Physical AI?

While automakers focus on industrial deployment, the education sector is building the workforce that will design and maintain these systems. Faraday Future, a California-based company focused on embodied AI (EAI), just announced the official launch of its EAI-EDU Nationwide Replication at Scale Strategy, following the successful completion of its first robotics summer camp.

The summer camp, held in partnership with the Lynwood and El Segundo school districts, graduated its first cohort of students who completed hands-on robotics projects. The program validated Faraday Future's curriculum, educational products, and business model for scaling robotics education across K-12 schools nationwide. The company plans to expand through partnerships with education organizations like Sequoia Education and Triple I, making after-school robotics programs a priority in public and private schools.

The educational approach goes beyond teaching AI alone. Faraday Future's curriculum integrates mechanical engineering, electrical hardware, complete robotics engineering, hands-on projects, and cross-disciplinary innovation skills. The goal is to transform robotics education from an exclusive experience available to a few into a scalable capability accessible to more schools and families.

Steps to Integrate Physical AI Education Into Schools

• Curriculum Development: Schools can adopt integrated programs that combine AI theory with mechanical engineering, electrical systems, and hands-on robotics projects rather than teaching AI in isolation.
• Partnership Models: Educational institutions can partner with robotics companies and manufacturers to access both curriculum materials and physical robot platforms, reducing the cost of entry.
• Real-World Application Scenarios: Programs should include practical projects tied to actual industry use cases, such as factory automation or retail environments, so students understand how robots solve real problems.
• Data Collection and Iteration: Schools can participate in continuous improvement cycles where student-operated robots generate data that improves the systems, creating a feedback loop between education and product development.

What Makes 24/7 Robot Interaction a Technological Breakthrough?

Beyond manufacturing and education, a startup called Richtech Robotics is experimenting with a novel approach to human-robot interaction: a humanoid robot called ADAM that streams live 24/7 on the internet, allowing global users to interact with it in real time. While this might sound like a gimmick, the underlying technology represents a meaningful shift in how embodied AI systems are developed and tested.

Most robots are demonstrated in controlled environments or deployed in isolated settings. ADAM, by contrast, must handle continuous, unpredictable user input from a global audience while coordinating natural language processing, physical movement, sensor interpretation, and real-time decision-making simultaneously. This creates a persistent feedback loop between human input and robotic behavior that is both synchronous and live.

The technical foundation for ADAM's capabilities relies on two key components. First, the robot uses NVIDIA Jetson Thor, an onboard computing system that processes AI models directly on the robot rather than relying entirely on cloud servers. This reduces latency, allowing near-instant responses to user commands. In robotics, even a fraction-of-a-second delay can disrupt coordinated movement or undermine the effectiveness of human-machine interaction. Second, ADAM runs on the NVIDIA Isaac robotics platform, a standardized development environment that enables rapid iteration and deployment of AI-driven updates.

The persistent operational model also serves a learning function. Unlike static training datasets, live interactions provide dynamic, context-rich data that can enhance the robot's conversational accuracy, behavioral adaptability, and contextual awareness. However, maintaining consistent performance under continuous, unpredictable input is a non-trivial engineering challenge that requires robust error handling and system reliability.

Richtech Robotics has positioned ADAM as a potential "robot influencer," merging social media engagement with physical robotic presence and AI interaction into a unified platform. The company is already exploring applications across hospitality, manufacturing, and automotive environments, suggesting that persistent, interactive robot platforms may become a standard approach for developing and evaluating embodied AI systems.

What Do These Developments Mean for the Physical AI Timeline?

The convergence of manufacturing scale-up, educational infrastructure, and persistent human-robot interaction suggests that physical AI is transitioning from the research phase into practical deployment. Xpeng's IRON entering mass production is comparable to Tesla's launch of the P7 vehicle with advanced autonomous driving features; it marks the beginning of technological emergence, not full maturity. The robots still need time to reach their full potential, but the trajectory is clear.

Education initiatives like Faraday Future's nationwide program indicate that the industry recognizes a critical bottleneck: the shortage of engineers and technicians trained to work with embodied AI systems. By building the curriculum and partnerships now, companies are preparing for a future where humanoid robots are commonplace in factories, retail stores, and other real-world environments.

The 24/7 interactive robot model pioneered by Richtech Robotics suggests another shift: from isolated demonstrations to persistent, scalable systems that improve through continuous real-world interaction. This approach aligns with broader trends in AI development, where systems learn and improve through exposure to diverse, unpredictable inputs rather than controlled laboratory conditions.

Together, these three developments paint a picture of an industry moving beyond hype toward infrastructure, workforce development, and operational deployment. Physical AI is no longer a future technology; it is becoming a present-day engineering and educational challenge.