Humanoid robot startup Figure raises $1B+ at $39B valuation – News

Figure AI Inc. today announced that it raised a funding round worth more than $1 billion from a group of prominent investors.

Parkway Venture Capital led the Series C deal. It was joined by Nvidia Corp., Intel Capital, Qualcomm Ventures, Salesforce Inc., LG Technology Ventures, T-Mobile Ventures and other institutional backers. Figure AI is now worth $39 billion, 15 times more than the valuation it received following its previous funding round last year.

Figure AI develops humanoid robots that can automate repetitive work in factories and perform household chores. The company’s flagship system, Figure 02, debuted last year. Figure says that the robot’s hands have “human-equivalent strength” with 16 degrees of freedom, a metric that measures how many motions they can perform. A human hand has 27 degrees of freedom.

In March, Figure detailed that it had completed the development of a new robot called the Figure 03. The company designed “almost the entire robot from scratch” including the actuators and motors it uses to move. Figure 03 also includes a custom battery pack with 2.3kWh of capacity, which allows the robot to operate for up to five hours per charge.

The company plans to manufacture the system at a factory called BotQ that it revealed in March. Figure will use the funding round announced today to expand the facility’s production capacity.

Its initial goal for BotQ is to make 12,000 robots per year. In the long term, Figure plans to grow that number “considerably.” The company will use some of the robots it will manufacture to automate BotQ’s production lines.

The factory uses custom software to coordinate the manufacturing workflow, measure production efficiency and track parts. It’s equipped with more advanced machinery than the facilities Figure relied on until now. According to the company, parts that previously took over a week to produce can be manufactured in about 20 seconds at BotQ.

Figure will also use its new funding to enhance Helix, the artificial intelligence model that powers its robots. Helix is a so-called VLA, or vision-language-action, model. It can analyze visual data from a Figure robot’s sensors, take actions based on that data and process natural language instructions from users.

Teaching a robot to interact with a new object usually requires manual programming. Figure says that a Helix-equipped machine can pick up certain household items even if it hasn’t encountered them before. Additionally, the AI model enables two robots to collaborate with one one another on tasks such as putting away the groceries.

The funding round will enable Figure to build new AI training infrastructure powered by Nvidia Corp. graphics cards. According to the company, its engineers will use the hardware to enhance Helix’s perception and reasoning skills. Figure also plans to enhance the multimodal datasets it uses to train the model.

Image: Figure