AI Advances Catapult Humanoid Robotics into the Mainstream

Markets were rocked in late January when the Chinese artificial intelligence firm, DeepSeek, appeared to unveil an artificial intelligence (AI) model that outperforms OpenAI’s for just $5.6 million. Western AI labs have been dropping billions on AI development, and preliminary analysis lends credence to DeepSeek claims.

Jensen Huang of Nvidia proclaimed at the 2025 Consumer Electronics Show (CES) that the world is entering the “ChatGPT moment” for humanoid robotics and autonomous systems.1 Highly dependent on sophisticated AI, if DeepSeek claims are legit, the future of humanoid robotics at an accessible price point are closer than any of us expected just one month ago. It might be a loss for Western AI development, but a certain win for incorporating humanoid robotics into the mainstream.

In a world of mind-boggling tech advances, the notion of a humanoid robot capable of performing useful tasks is somewhat antiquated. After all, the popular cartoon “The Jetsons” had already predicted the future over 40 years ago with robots buzzing about, taking care of everything humans would rather delegate.

Yet, it turns out engineering robots to take on menial or repetitive tasks is not all that easy. But, the convergence of AI, declining robotics hardware costs, and a better understanding of operations systems are finally advancing the humanoid robotics market toward mainstream adoption.

Humanoid robots combine mechanical systems (mechatronics) and AI, both of which have developed at breakneck speeds over the last decade. MIT, Stanford, Carnegie Mellon, and a host of peer universities offer robotics and AI engineering degrees centered on riding the emerging growth wave, and humanoid robotics has been actively deployed across a range of industries.

Competition in the humanoid robotics sector is fierce. While a budding DeepSeek partnership is likely in the works, it will need to catch up with some industry heavyweights.

In February 2024, OpenAI announced a partnership with the California-based robotics startup, Figure.2 Figure initially raised $70 million in Series A funding in 2023 and closed $675 million in Series B last year.3 By late December 2024, CEO Brett Adcock took to LinkedIn with news that every startup hopes to eventually share:

OpenAI CEO Sam Altman signaled in a podcast with Microsoft co-founder Bill Gates that the company is open to renewed robotics interest again.5 It has taken decades for robotics engineers to manufacture robots that can actually walk upright, and initial AI interest in robotics had waned as a result. But OpenAI restarted its robotics research group and Figure is in talks with BMW to deploy the 02 robot for handling fitted sheet-metal parts on its Spartanburg, SC production line.

Another humanoid robotics upstart making waves is Apptronik, home to Apollo and its novel use of linear actuators.

Normally found in medical equipment such as hospital beds or adjustable furniture, actuators employed in robotics allow robots to perform complex tasks with extremely high levels of accuracy. Rather than using gear trains and rotary motors as is typical, Apptronik and Apollo earned a 2024 Robotics Innovation Award for their novel actuator use.

Like OpenAI with Figure, the Google DeepMind robotics team is now working with Apptronik to eventually enable reasoning and action in real-world applications.7 Apptronik’s long-term objective is humanoid robotics serving in healthcare and performing basic functions around the home. DeepMind has a wealth of machine learning, engineering, and physics simulation experience, and the two are advancing rapidly to solve an emerging need for more healthcare support as the US and most developing nations deal with an aging population.

“Mega,” as it sounds, is a wide-reaching initiative created by Nvidia to optimize and test humanoid robotic fleets. One of the greatest challenges humanoid robotics faces is data scarcity. Large language models (LLMs) leverage vast amounts of textual data from the Internet, but robots require interaction with their environment to learn.8 Nvidia’s “Mega” omniverse blueprint is a virtual replica of a physical environment - a digital twin - where designers can tinker with infinite scenarios and populate themselves with data.9

In line with “Mega” is yet another Nvidia product - the Isaac GROOT Blueprint. This technology aids developers in generating motion data to train humanoid robots via imitation learning.10 The robots acquire new skills by watching and mimicking the human demonstrations. Apple Vision Pro also aids in capturing human actions through a digital twin, thus expanding the dataset and advancing robot experimentation safely in an open-source, digital environment.

Few folks, with the exception of elected President Donald Trump, are as searched today as Elon Musk. The world’s richest person, with a total net worth of $450 billion, absolutely dwarves the second person on the list - Jeff Bezos at $242 billion.11 Musk will now be working actively with the Trump administration, and like Figure and Apptronik, Musk’s humanoid robot Optimus has also found an AI partner.

In the case of Optimus, the fortunate AI partner is part of the Musk family - his AI startup, xAI. Last May, xAI raised $6 billion from Sequoia Capital, Saudi Prince al Waleed bin Talal, and Andreessen Horowitz, among others.12 Musk is extremely bullish on the future of humanoid robotics, catching every attendee’s attention in October at the Future Investment Initiative Conference with this prognostication:

I think by 2040, probably there are more humanoid robots than there are people. Every country will have an AI or multiple AIs, and there will be a lot of robots, way more robots than people.13

The latest generation of Optimus is a full-blown bipedal machine that can work autonomously in Tesla factories. The biggest challenge at the moment is the cost. Musk wants to get the entry-level price under $30,000, but most humanoid robots are still north of $50,000. DeepSeek advances could eventually shatter expectations.

Humanoid robotics can assist with jobs that are either dangerous or can easily be automized.

Governments will invariably be faced with pressure to prevent labor displacement with entrenched labor unions leading the charge. It remains to be seen what the political appetite is for mass incorporation of humanoid robots into key sectors.

Meanwhile, Goldman Sachs projects near-term investment opportunities in the component makers that are part of the larger humanoid robotics supply chain. A labor substitution rate for dangerous jobs and car manufacturing of 5-15% would yield demand for up to 3.5 million humanoid robots globally.14

Sources:

1. Lee, Ricki. January 9, 2025. “CES 2025: Nvidia CEO unveils robotics platform and agentic AI advances.” Tech Informed.
2. O’Brien, Matt. February 29, 2024. “Humanoid robot-maker Figure partners with OpenAI and gets backing from tech giants.” Associated Press.
3. Demaitre, Eugene. December 29, 2024. “Figure AI ships Figure O2 humanoid robots to a paying customer.” The Robot Report.
4. Adcock, Brett. December 22, 2024. LinkedIn Post.
5. Gates, Bill. January 11, 2024. “My conversation with Sam Altman.” Unconfuse Me With Bill Gates.
6. Demaitre, Eugene. December 19, 2024. “Apptronik partners with Google DeepMind to advance humanoid robots with AI.” The Robot Report.
7. Idem.
8. Robotics Biz. Editorial. February 29, 2024. “Can Robotics Overcome Its Data Scarcity Challenge?” Robotics Biz.
9. Huang, Madison. January 6, 2025. “NVIDIA Unveils ‘Mega’ Omniverse Blueprint for Building industrial Robot Fleet Digital Twins.” NVIDIA.
10. Huang, Spencer. January 6, 2025. “NVIDIA Announces Isaac GROOT Blueprint to Accelerate Humanoid Robotics Development.” NVIDIA.
11. Bloomberg. January 16, 2025. “Bloomberg Billionaires Index.”
12. The Associated Press. May 27, 2024. “Elon Musk’s xAI raises $6 billion from Sequoia, Andreessen, and Saudi royals as it hails ´significant strides´ in AI research.”
13. Klender, Joey. October 30, 2024. “Tesla CEO Elon Musk makes bold claim that would be huge for Optimus.” Teslarati.
14. Source: Goldman Sachs. February 27, 2024. “The global market for humanoid robots could reach $38 billion by 2035.”