Key Takeaways
- Signaloid will unveil its C0‑ASIC for physical AI at Bosch Connected World (June 10‑11 2026, Berlin), projecting up to 1,000× better performance‑per‑Watt than current solutions.
- The company’s existing distribution‑extended compute hardware (UxHw®) already delivers >37‑fold speedups for robotics sensor‑fusion algorithms and >600‑fold gains for infrared‑sensor data analysis.
- UxHw treats numbers as probability distributions, enabling a single execution to replace millions of iterative calculations used by conventional GPUs or CPUs.
- The C0‑ASIC will complement Signaloid’s edge hardware modules and virtualization‑based offerings, easing integration with industrial platforms such as Bosch Rexroth’s ctrlX core X2 and X3 PLCs.
- Founded by former Cambridge professor Phillip Stanley‑Marbell (ex‑Bell Labs, IBM, Apple, MIT), Signaloid serves >3,000 users worldwide via cloud, on‑premises, and low‑power edge deployments.
Signaloid’s Upcoming ASIC Preview at Bosch Connected World
British AI hardware firm Signaloid announced that it will preview its newly taped‑out C0‑ASIC for physical AI at Bosch Connected World, scheduled for June 10‑11 2026 in Berlin. The event will showcase the ASIC’s potential to revolutionize compute‑intensive workloads in robotics, industrial automation, and probabilistic AI. By highlighting the chip at a major industrial technology conference, Signaloid aims to demonstrate real‑world relevance and attract partners seeking ultra‑efficient edge computing solutions.
Performance Promises of the C0‑ASIC
According to the company, the C0‑ASIC is projected to deliver up to 1,000× better performance‑per‑Watt compared with today’s state‑of‑the‑art approaches for the same workloads. This metric combines raw throughput with energy efficiency, a critical factor for battery‑powered autonomous mobile robots (AMRs) and continuously operating factory equipment. If realized, the gain would allow complex probabilistic algorithms to run on modest power budgets, opening doors to longer‑range mobile robots and always‑on predictive maintenance systems without prohibitive energy costs.
Why Physical AI Demands a Different Compute Model
Many robotics and AI algorithms must evaluate hundreds of thousands—or even millions—of possible scenarios each second to estimate a robot’s pose, track a moving drone, or predict equipment failure. Because these scenarios are not equally likely, traditional processors approximate the ideal answer by repeatedly sampling and averaging, which consumes considerable time and energy. Physical AI therefore benefits from hardware that can inherently handle uncertainty and probability, eliminating the need for massive brute‑force iteration.
Signaloid’s Distribution‑Extended Compute (UxHw®)
Signaloid’s core innovation, UxHw, represents values not as single numbers but as arbitrary non‑uniform ranges—essentially probability distributions—and performs computation directly on this digital form. A single execution of existing software on a UxHw‑enabled platform can therefore produce results that would otherwise require millions of iterative steps on conventional CPUs or GPUs. In head‑to‑head benchmarks against the latest high‑end computing platforms, UxHw already achieves 1,000‑fold speedups, with further improvements anticipated from the forthcoming C0‑ASIC.
Real‑World Speedups Already Demonstrated
Even before the C0‑ASIC’s release, Signaloid’s technology has shown impressive gains. Cloud‑ and FPGA‑based implementations of UxHw deliver over 600‑fold speedups for infrared sensor data analysis and more than 37‑fold acceleration for particle‑filter sensor‑fusion algorithms—core techniques used in localization and mapping for autonomous robots. These results validate the approach’s applicability to demanding perception pipelines and suggest that the ASIC will push performance even higher.
How the C0‑ASIC Complements Existing Offerings
The C0‑ASIC is designed to work alongside Signaloid’s current family of UxHw hardware modules and its virtualization‑binary‑translation solution. This modular strategy lets customers adopt the ASIC where maximum efficiency is needed while retaining flexibility for software‑only or FPGA‑based deployments. Notably, the modules are already earmarked for integration with Bosch Rexroth’s ctrlX core X2 and X3 programmable logic controllers (PLCs), indicating a clear pathway to factory‑floor adoption.
Broader Application Scope Cited by Leadership
Phillip Stanley‑Marbell, Signaloid’s founder and CEO, emphasized that the compute workloads targeted by UxHw extend beyond physical AI and robotics. He noted relevance to supply‑chain modeling, logistics, quantitative finance, and any domain where probabilistic reasoning is central. By framing many computationally challenging problems as distributions‑processing tasks, Signaloid aims to create a unified hardware platform that serves multiple high‑value industries.
About Signaloid: Founder, History, and Reach
Signaloid was founded by Prof. Phillip Stanley‑Marbell, a former Professor of Physical Computation at the University of Cambridge whose career includes research stints at Bell Labs, IBM, Apple, and MIT. The company provides a computing platform that accelerates workloads reformulated to process probability distributions. Today, more than 3,000 users worldwide rely on Signaloid’s technology, which is available as cloud services, on‑premises appliances, and low‑power edge hardware. Further information can be found at www.signaloid.com, and press inquiries should be directed to [email protected].
This summary captures the essential announcements, technical distinctions, performance claims, and contextual details from the original source while adhering to the requested length, structure, and formatting.