Robotics pioneer Cyberdyne looks to Silicon Valley for “artificial brain” partnerships

The Japanese technology group Cyberdyne, a pioneer in wearable exoskeleton devices, has teamed up with Silicon Valley’s Pegasus Tech Ventures to launch a ¥10bn yen ($62m) corporate venture capital fund. The vehicle, which will be managed through Pegasus’s venture-capital-as-a-service platform, marks a strategic pivot for established robotics firms that are racing to integrate generative artificial intelligence into their physical hardware.

For decades, the robotics industry was defined by mechanical excellence that far outpaced the software required to control it. As Anis Uzzaman, founder and chief executive of Pegasus Tech Ventures, notes, the arrival of large language models has fundamentally altered the landscape.

“2026 is all about physical AI,” says Uzzaman, whose firm has backed heavyweights such as OpenAI, Anthropic and Figure AI.

The missing link in machine intelligence

The partnership highlights a growing anxiety among traditional robotics players that they might “miss the boat” on the AI revolution. Cyberdyne, founded in 2004 by Professor Yoshiyuki Sankai, became famous for HAL (Hybrid Assistive Limb), an exoskeleton designed to aid human mobility by responding to bioelectric signals. However, even the most advanced hardware has historically struggled with the nuance of human interaction.

This challenge is perhaps best illustrated by the fate of Honda’s famous humanoid robot, Asimo. Despite 20 years of research and a high-profile demonstration playing football with President Obama, the project was eventually shuttered because the “software was behind” the mechanical and electrical infrastructure. Generative AI has provided the “light” that legacy manufacturers have been waiting for, offering a way to move beyond pre-programmed routines towards autonomous reasoning.

A new frontier for industrial investment

Investors are responding to this shift with significant capital. Beyond manufacturing and warehouses, where robots such as Agility’s Digit are already helping with logistics, the focus is shifting toward “human assistive” roles. In Japan, where a shrinking labour force and an ageing population have created a critical shortage of healthcare workers, physical AI is seen as a strategic necessity rather than a luxury.

“Physical AI is the right space that Cyberdyne can get into,” says Uzzaman. He notes that legacy companies are desperate to bridge the “cultural barrier” between traditional engineering and the fast-moving AI labs of California. By launching a dedicated venture arm, Cyberdyne aims to secure early relationships with the next generation of AI “brains” that will eventually power its “limbs”.

The race for strategic partnerships

This trend is not confined to Japan. European automotive giants are increasingly looking to startups to provide the intelligence for their factory floors; Agility Robotics has partnered with BMW, while Apptronik is collaborating with Mercedes-Benz. The Pegasus-Cyberdyne fund is designed to facilitate similar joint ventures and licensing deals.

As the sector matures, the investment models are also becoming more sophisticated. To compete for access to the inner circle of Silicon Valley, Pegasus often uses a hybrid model, combining smaller checks from different sectors—such as automotive, healthcare, and gaming—to build larger, more competitive investment positions in firms like OpenAI.

As established players seek to modernise, they are finding that capital alone is not enough. They require the technical “transfer function” that can translate a Silicon Valley algorithm into a reliable industrial tool. For Cyberdyne and its peers, the goal is clear: to ensure that the robots of the next decade have the intelligence to match their mechanical strength.