As robots, drones and autonomous machines become more capable, the chips powering them face a growing challenge: they must interpret the physical world, run artificial intelligence and control movement in real time. Semiconductor startup Ubitium is developing a processor designed to bring these tasks together on a single architecture. The company has raised $14 million in new seed funding to move its technology closer to commercialisation and address the computing demands of the next generation of intelligent machines.
The funding round was led by Ibex Investors, with participation from OpenOcean, existing investors KBC Focus Fund and Runa Capital, and Inflection, as well as Florian Seibel, co-founder and co-CEO of Quantum Systems, who invested personally. Founded in 2024, Ubitium has now raised a total of $18 million in seed funding.
Rethinking How Intelligent Machines Compute
Many autonomous systems rely on several specialised processors to perform different tasks. One chip may handle sensor signals, another may run AI inference, while a separate processor manages real-time control. Although this division allows each chip to focus on a particular function, it can also increase system complexity, cost and the amount of data transferred between components.
Ubitium aims to simplify this arrangement through a reconfigurable processor that combines sensing, AI inference and real time control within a unified architecture. Its programmable processing elements can dynamically allocate computing resources according to a machine’s workload.
The approach is designed for physical AI applications, including robotics, drones, vehicles and industrial equipment, where machines must continuously process information from their surroundings and respond within strict timing requirements.
One Architecture for Multiple Tasks
Ubitium’s technology combines unified memory with a reconfigurable processing array. Instead of relying on separate specialised chips for different workloads, manufacturers could use a common processor architecture to handle multiple computing functions.
The company also uses standard RISC V development tools, which are intended to make the platform easier for developers to work with. A shared software toolchain could reduce the effort involved in building and maintaining systems that otherwise depend on several processor types.
By reducing data movement between chips and consolidating computing functions, Ubitium aims to simplify hardware design and lower the overall cost of developing intelligent machines.
First Chip Demonstrates the Approach
Ubitium has already developed its first chip, the UB108, which contains 256 processing elements and was manufactured using Samsung’s 8 nanometre process.
The company says the chip has demonstrated signal processing, AI inference and real time control running on the same reconfigurable array. This milestone provides an initial demonstration of its central proposition: that multiple computing workloads used in physical AI can operate within one processor architecture.
The achievement also marks a step towards turning the company’s technology from a development project into a commercial product.
UB410 Moves Towards Commercialisation
The next major milestone is the UB410, Ubitium’s planned commercial processor featuring 4,096 processing elements. Initial applications will focus on high-resolution radar and audio systems, while the broader architecture is intended to support autonomous systems such as drones.
CEO and co-founder Hyun Shin Cho has argued that inference is only one part of the physical AI computing challenge. Machines must also interpret real-world signals and control their actions, tasks that are frequently handled by separate chips today.
Ubitium says its first chip demonstrated the combined approach within a year, and the latest funding will help advance the UB410 towards commercial readiness.
Funding the Next Phase
The company plans to use the new capital to develop the UB410, with tape-out and delivery of evaluation units targeted for 2027. Tape out is the stage at which a chip design is finalised for manufacturing.
If the processor meets its goals, Ubitium could offer manufacturers an alternative to systems built around multiple specialised chips. The company’s next challenge will be to demonstrate the architecture’s performance, efficiency and practical advantages across real-world workloads as it moves towards commercial deployment.
