Is ‘physical AI’ the next growth phase in the AI cycle? How to get ahead

The advent of so-called physical artificial intelligence is believed by many to be the next frontier of the AI investing cycle, opening up opportunities to get behind robotics makers, auto suppliers and specialty semiconductor companies. Investors are looking for another way to play AI after huge runs in Nvidia and the other stocks powering and producing digital bots like ChatGPT. Search bots were the start and the technology has now moved into “agentic AI” like customer service representatives and research assistants. Now the focus for investors analyzing the leading edge of this technology is becoming interaction in the physical world with devices acting independently through artificial intelligence, like robot nurses and self-driving drones and cars. And analysts believe this is where the big money will be. “As you go into 2025, agentic AI is that next inflection point here before you hit that physical AI moment … like with everything else in the world, you have to crawl before you can walk and then run,” said CFRA senior equity analyst Angelo Zino. “Gen AI hasn’t really seen much monetization thus far. So when you start thinking about agentic AI and robotics and what have you, there’s a real kind of belief out there — and expectation. You’ll see those use cases start to come about this year.” Uber , along with Mobileye , a maker of driver assistance tech, and Aptiv , an auto parts supplier, could be beneficiaries of this futuristic trend. The burgeoning humanoid robot market also has several big-name players, which we previously reported on, like Tesla and XPeng . And of course, Nvidia is a huge frontrunner in physical AI as well, according to several Wall Street analysts. Wall Street is expecting realistic AI-enabled agent applications to start coming out this year and physical AI innovations to begin playing out later, most likely in self-driving cars first . Physical AI has created “an additional leg to the stool” to the AI build-out phase, said Goldman Sachs analyst Toshiya Hari. “There’s an additional driver now that augments what’s going on today and AI broadly.” Nvidia and the ‘chatGPT moment’ for robotics Nvidia has been the AI darling for as long as investors can remember, as it owns the majority of the AI chips market and continues to be a standout name in the agentic AI space. But its efforts in robotics haven’t yet caught up to Wall Street’s near-term estimates on the stock. Nvidia founder and CEO Jensen Huang’s Jan. 6 CES keynote left investors and analysts optimistic on the company’s announcements across its hardware and software products. Huang touted during his speech that “the ChatGPT moment for general robotics is just around the corner,” and said the company expects to enable the development of three key robots in the coming years: agentic agents, self-driving vehicles and humanoid robots. Huang spent much the keynote on Nvidia’s development of agentic models that can perform tasks like coding and chat support. He also announced Nvidia’s Cosmos platform for companies looking to train and develop physical AI systems. The Cosmos world foundation models can generate physics-based videos from several inputs, like video, image text, as well as robot sensor and motion data, according to a recent Nvidia press release. Eminent robotics and automotive companies, like XPeng , ridesharing giant Uber and Agile Robots, have already adopted the platform, the company has said. Several humanoid makers have also adopted Nvidia’s Project GR00T model. In an interview, Goldman’s Hari said he was “struck” by the Cosmos announcement and the pace at which Nvidia is innovating. His buy rating and $165 price target are driven by Nvidia’s Blackwell GPU sales and generative AI-fueled data center growth, as well as continued cloud and enterprise spending, which he expects to see throughout 2025 and 2026. Although physical AI is not a big part of Hari’s Nvidia bull thesis right now, the analyst said that the “concept of physical AI would lead to increased confidence in this business growing year-over-year” in his 2027 and 2028 estimates. CFRA’s Zino and Bank of America senior analyst Vivek Arya also remain bullish on Nvidia’s broader footprint in AI, especially after Huang’s keynote. But they also reiterated that multiple unknowns exist when it comes to how long the company’s robotics efforts can translate into its financials. Arya said that Nvidia’s approach to being the brains behind physical AI, rather than the maker of actual autonomous cars and humanoids, puts the company in a spot “where the big pools of profit are” given that consumer device makers ultimately deal with lower margins and greater competition. Zino said the Nvidia’s existing swathe of customers and partners should eventually allow it to “continue to really gain a lot of traction as physical AI becomes more of an opportunity.” He added that “it’s too early to really say how this is all going to evolve, but Nvidia is going to clearly be a winner here.” Other emerging winners of ‘physical AI’ Several companies involved in the technologies powering physical AI systems, like sensors and control system makers, also stand to benefit from the trend. Oppenheimer analyst Colin Rusch named autonomous driving tech developers Aptiv and Mobileye, warehouse automation company Symbotic and high-energy lithium-ion battery maker Amprius Technologies as some top picks. “We see physical AI as a nascent interdisciplinary field poised to transform the industrial complex through increased asset productivity and labor efficiency,” Rusch said in a recent note. “We see outsized investment return potential in sensing, sensor fusion, compute hardware & software architecture, connectivity, controls, and AI training strategies,” he continued, adding that he expects the maturing technology to benefit companies with strong balance sheets and access to incremental investment capital. Auto parts supplier Aptiv is a strong play in sensors, as the company provides radar and sensor fusion technologies that help build “a truly comprehensive and reliable environmental model,” the analyst said. Sensor fusion technologies take a combination of data from radar, lidar and camera sensors to create a detailed image of the driving environment. Intel-owned Mobileye is also a top Oppenheimer pick in the software architecture side of physical AI. “While select vertically-integrated players will likely emerge in an autonomous future, we believe MBLY will be the key enabler of most legacy OEMs delivering autonomous functionality,” Oppenheimer stated in a note. Rusch noted Mobileye’s “leading” expertise in sensors and a Compound-AI system approach for self-driving systems. MBLY 1Y mountain Mobileye stock performance over the past year. Amprius, another top physical AI pick from Oppenheimer, has seen its shares pop roughly 105% over the past month fueled by the company’s Tuesday announcement of its SiCore cell, which is part of a platform that provides high-performance energy and power performance for drones, electric vehicle and aviation applications. The firm’s price target suggests a whopping rally of more than 316% over the next year. There’s no exact timeline on when the physical AI opportunity will emerge and most analysts are still struggling to incorporate it in to their short- and medium-term models. “Most of these companies, most of these great minds a year ago were talking about hitting that point within a three-year span or maybe sooner, others a little bit later,” Zino said. That’s “when we start seeing a multi-trillion dollar TAM opportunity on the physical AI side of things.”