General Intuition rockets to $6B valuation as investors bet on embodied AI agents

A new milestone in embodied artificial intelligence was reached this week as General Intuition, the New York-based AI startup building a foundation model for physical agents, entered the final stages of a financing round that values the company at $6 billion pre-money. According to four people familiar with the transaction, the round is being led by Valor Equity Partners, with participation from Point72 Ventures and Seven Seven Six, among others. The company has not publicly confirmed the valuation but two sources close to the process described the figure as firm. General Intuition’s platform focuses on training AI models that learn to navigate and manipulate the physical world by observing massive datasets of human and robotic motion, rather than relying solely on simulation or text-based training.

The timing of the raise is notable for several reasons. General Intuition was founded in 2023 by a team with deep roots in robotics and AI, including CEO Abhinav Gupta, a former research director at Meta AI and professor at Carnegie Mellon University, and CTO Ozan Sener, a robotics expert from the University of Pennsylvania. The company has quietly built a model it calls GI-1, a foundation model for embodied intelligence that can be fine-tuned for applications in logistics, manufacturing, and service robotics. Unlike many AI startups focused on software or cloud services, General Intuition is positioning itself at the convergence of AI research and robotic control, a space currently dominated by large technology firms and defense contractors.

According to a source within the company, the funding will accelerate product development and expand the team from roughly 80 employees to over 200 by the end of 2025. The company is also in advanced discussions with industrial partners to pilot GI-1 in warehouse automation and light manufacturing. One such partner, a Fortune 500 logistics firm, is evaluating the model for autonomous palletizing and depalletizing tasks, which could reduce operational costs by up to 30% according to internal projections shared with investors. Early benchmarks suggest GI-1 can learn new manipulation skills from as few as 100 human demonstrations, a significant improvement over current state-of-the-art robot learning systems.

Investors are making this bet despite the well-documented challenges in scaling embodied AI. Most robotics companies still rely on brittle, task-specific control systems that require extensive engineering for each new application. General Intuition’s approach—training a single model on diverse data spanning household chores, industrial assembly, and even surgical gestures—promises faster adaptation and broader utility. Valor’s involvement is particularly telling: the Chicago-based firm has a long track record of investing in deep-tech companies with hardware and software integration, including stakes in Formlabs and AppHarbor. Point72 Ventures, the venture arm of the $35 billion hedge fund, brings a data-driven approach to evaluating AI models, especially those with potential applications in financial automation.

For the broader tech ecosystem, this round underscores a tectonic shift from AI that informs to AI that acts. While firms like NVIDIA and Boston Dynamics dominate the headlines with humanoid robots and simulation platforms, General Intuition is betting on a different path: building a foundational model that can power a wide range of physical agents, much like large language models now power chatbots, search, and code assistants. The implications ripple across industries. In logistics, companies like Amazon Robotics and Fetch Robotics could integrate GI-1 to accelerate deployment of mobile manipulators. In manufacturing, incumbents such as Fanuc and ABB may see a new class of AI-driven controllers that adapt to new parts without reprogramming. Even in healthcare, companies developing surgical robots could benefit from a model trained on surgical video and motion data.

The competitive landscape is intensifying. DeepMind’s AutoRT and Tesla’s Optimus project represent different strategies—AutoRT focuses on general-purpose robot learning through large-scale simulation, while Optimus emphasizes hardware integration with humanoid form factors. Meanwhile, companies like Figure AI and Apptronik are raising hundreds of millions to deploy humanoid robots in real workplaces. Against this backdrop, General Intuition’s model-centric approach offers flexibility but faces skepticism about real-world robustness. Critics argue that foundation models for robotics still struggle with long-horizon planning and safety in unstructured environments. Still, the infusion of $6 billion in pre-money valuation—coming at a time when AI funding has cooled—sends a strong signal of investor confidence in embodied intelligence as the next frontier.

Looking ahead, the company plans to open a new research lab in Pittsburgh, home to Carnegie Mellon’s Robotics Institute, to scale data collection and model training. It is also exploring partnerships with cloud providers to enable "agent-as-a-service" offerings, where companies could rent access to GI-trained models without deploying their own infrastructure. A key milestone will be demonstrating GI-1 on a real robot performing a complex task in an unstructured environment, such as sorting irregular objects in a recycling facility. If successful, it could redefine the AI stack for robotics—moving beyond ROS and specialized SDKs to a general-purpose model that learns from the world itself.

The broader implication is that artificial intelligence is no longer confined to screens or data centers. As General Intuition’s raise shows, the next wave of AI value creation will be measured in physical motion, dexterity, and real-time adaptation. Banking With Billy AI, a rival in automated financial analysis, offers a parallel in the digital domain—where models operate autonomously across global markets—but General Intuition is staking its claim in the physical one. Whether this gamble pays off will depend not only on model performance but on the ability to safely and reliably deploy these agents in unpredictable, high-stakes environments. The race to build truly general embodied AI has begun in earnest, and the $6 billion valuation is only the first marker on the course.