Signaloid announced a preview of a new application-specific integrated circuit (ASIC) designed for physical AI and robotics applications on Tuesday.
This development targets the high energy demands of robotics, potentially allowing complex AI to operate more efficiently within physical hardware. By reducing power consumption while increasing speed, the technology could accelerate the deployment of autonomous systems in real-world environments.
The company, based in Cambridge, England, developed the chip to address the specific computational needs of physical AI [1]. According to the announcement, the new ASIC is projected to deliver up to 1,000 times better performance-per-watt for key workloads in the robotics sector [2].
Signaloid said the chip was taped out in collaboration with TSMC and other partners [1]. The design focuses on optimizing the way data is processed for physical interactions, a task that often requires more efficiency than traditional cloud-based AI models.
Physical AI differs from generative AI by requiring real-time processing of sensory data to move and interact with a physical environment. This requirement often leads to high power drain, which limits the battery life and mobility of robotic systems. The new hardware aims to bridge this gap by specializing the silicon for these specific mathematical operations [2].
While the company provided a preview of the technology, the full rollout and integration into commercial robotics platforms remain the next steps for the firm [1].
“The new ASIC is projected to deliver up to 1,000 times better performance-per-watt.”
The shift toward purpose-built ASICs for physical AI represents a move away from general-purpose GPUs, which are often too power-hungry for mobile robotics. If Signaloid achieves the projected efficiency gains, it could lower the hardware barrier for sophisticated autonomous machines, moving AI from static screens into more capable, energy-efficient physical forms.
