Ubitium raises USD $14 million to build one AI chip

Ubitium has raised USD $14 million in seed funding, bringing the German semiconductor startup's total seed funding to USD $18 million.

Ibex Investors led the round, with participation from OpenOcean, existing investors KBC Focus Fund, Runa Capital and Inflection, and a personal investment from Florian Seibel, Co-Founder and Co-Chief Executive Officer of Quantum Systems.

The funding comes as Ubitium develops a single processor for autonomous machines that currently rely on several types of chips to handle sensing, artificial intelligence and control tasks. Its first silicon has already been produced and tested.

Autonomous machines such as drones, robots, vehicles and industrial equipment often split workloads across a digital signal processor for sensing, a neural processing unit for AI and a central processing unit for control. That can add engineering complexity and increase data movement between chips, particularly in systems that must operate in real time and within power limits.

Ubitium's approach is to use one processor architecture for all three functions. Its design is based on an array of programmable elements that can be assigned to different tasks at runtime, depending on the machine's workload.

Its first chip, called the UB108, contains 256 processing elements and was manufactured using Samsung's 8nm process. The chip demonstrated signal processing, AI inference and real-time control on the same reconfigurable array, according to the company.

The next planned processor, the UB410, is intended to have 4,096 processing elements. Ubitium plans to deliver evaluation units in 2027.

Market focus

The startup said its initial customers are building radar and audio systems, where sensing workloads can require both signal processing and AI on embedded hardware. It is also targeting broader autonomous systems, including drones, where sensing, decision-making and control must be handled on board.

Ubitium also pointed to a wider market shift as AI systems move from software applications into machines that interact with the physical world. It cited McKinsey's base-case forecast that general-purpose robotics alone could become a USD $370 billion market by 2040.

Ubitium was founded in 2024 by Chief Executive Officer Hyun Shin Cho and Chief Technology Officer Martin Vorbach, who has worked in reconfigurable computing for more than three decades. The company is based in Düsseldorf.

It holds 17 granted patents related to its core technology, with further patents pending. The wider team has previous experience at Intel, Nvidia, Apple and Texas Instruments, according to the company.

Ubitium's architecture uses unified memory and standard RISC-V tools, which it says could replace a mix of specialised processors and software toolchains. The aim is to let manufacturers reuse one computing platform across different products rather than integrating separate chips for each function.

"Physical AI machines bring a whole new set of computing challenges, and the inference isn't the hardest one," said Hyun Shin Cho, Chief Executive Officer and Co-Founder of Ubitium.

"Sensing the real world and controlling the machine in it are just as hard, and today those still run on separate chips. We built one Physical AI processor that does all three, proved it in silicon in a year, and this funding turns it into a product."

Investors backing the round framed the bet around the need for autonomous systems to keep operating without relying on external communications links. That is particularly relevant in areas such as drones and other mobile machines, where limits on power, weight and onboard computing can shape system design.

"An autonomous machine operating in the real world must keep working when communications are unavailable. It has to process its surroundings, make decisions and control itself on board, within strict limits on power and weight," said Florian Seibel, Co-Founder and Co-Chief Executive Officer of Quantum Systems.

"Today that means integrating a stack of chips that were not designed to work together, and every one of them costs weight, power and integration time we don't have. Ubitium is building the one processor that does that whole job. That's why I invested."

Jeffrey Peters, Partner at Ibex Investors, said the firm viewed the company's reconfigurable design as a way to keep pace with rapid changes in AI workloads. OpenOcean made a similar case for new chip architectures as AI systems move into robots and other machines operating outside traditional computing environments.

"The pace of AI is only increasing, far exceeding the ability of hardware to keep up and manage," said Jeffrey Peters, Partner at Ibex Investors.

"Ubitium solves this with a universal reconfigurable architecture that can deploy the latest advances on a single piece of silicon with a single set of development tools. Ubitium is truly hardware at the pace of software."