The Nervous System of Tomorrow: Silicon's Role in Safe Robotics

SANTA CLARA, Calif. – November 09, 2026 – For the past decade, the technological world has been captivated by the evolution of artificial intelligence—its ability to process vast amounts of data, generate language, and recognize complex patterns. Yet, as we look toward a future where technology physically intersects with our daily lives, a critical realization has emerged: the ability to "think" is fundamentally different from the ability to "act." If humanoid robots and advanced industrial automation are to thrive in our communities, hospitals, and factories, they must bridge the gap between high-level cognitive processing and precise, secure physical movement.

Addressing this profound challenge, MIPS, an operating entity under GlobalFoundries (GF), today unveiled the MIPS Actus reference platform ahead of the electronica 2026 trade fair. Designed to act as the nervous system for the next generation of robotics, Actus tackles the most demanding engineering bottleneck in the physical AI landscape: the localized control of robotic joints.

"Robots will earn trust through the way they move autonomously and safely," said Sameer Wasson, CEO of MIPS. "Actus closes the gap between AI decisions and physical action by delivering deterministic control at every joint and a secure, time-sensitive network across the robot. Together with our Acies and Aegis platforms, it creates a complete physical AI foundation built on open, efficient technology and the manufacturing scale of GF."

Solving the 'Last Mile' of Robotic Movement

In the anatomy of a humanoid robot, the "last mile"—the space from the elbow to the fingertips—presents an agonizing challenge for hardware engineers. Each joint must execute incredibly smooth, human-like movements while operating under severe constraints regarding space, weight, power consumption, and thermal dissipation. Traditional architectures, which often rely on central processing units sending commands over lag-prone networks to "dumb" actuators, simply cannot deliver the dexterity required to hold an egg, assist a patient, or assemble delicate microelectronics.

Actus fundamentally redesigns this architecture by embedding deterministic motor control, in-robot networking, and edge inference onto a single chip located directly within the robot joint. Built on GlobalFoundries’ energy-efficient FDX FD-SOI (Fully Depleted Silicon On Insulator) technology, the platform is uniquely equipped to handle the thermal and power limitations of compact robotic limbs. FD-SOI technology is renowned for its ability to deliver high performance at ultra-low power, minimizing the heat that could otherwise degrade mechanical components or harm human operators.

At the computational heart of Actus are two MIPS Atlas M8500 real-time RISC-V processors. These processors utilize hardware multithreading, a design choice that effectively eliminates the context-switch penalties typically incurred during interrupt services. In practical terms, this allows the robot to react to sudden physical changes—like a slipping object or an unexpected human touch—in sub-10 microsecond loops.

Furthermore, MIPS has engineered a dedicated trigonometric acceleration unit that delivers more than 16 times the performance of standard trigonometric libraries on field-oriented control algorithms. Paired with 100-picosecond-class high-resolution pulse-width modulation (PWM) and sigma-delta analog-to-digital converters, this mathematical acceleration translates directly into the fluid, high-torque, and high-RPM movements that separate true humanoid dexterity from jerky, mechanical automation.

The Open Frontier: RISC-V Challenges the Status Quo

The introduction of Actus also signals a broader shift in the semiconductor landscape, highlighting the rapid maturation of the open-source RISC-V architecture in mission-critical industrial applications. For years, proprietary architectures from industry giants like Arm, Texas Instruments, and STMicroelectronics have dominated the embedded control market. However, as robotics manufacturers seek greater customization and freedom from licensing lock-ins, RISC-V has emerged as a formidable alternative.

"Robot makers are moving a new generation of completely novel designs toward production, and the actuators and networks connecting them now pose the toughest engineering problems," said Edward Wilford, Senior Research Director at Omdia. "Platforms that can combine real-time motor control with deterministic networking on a single chip will be addressing a clear gap in the market. MIPS is positioning Actus to fill it on an open RISC-V foundation."

By leveraging the modularity of RISC-V, MIPS has been able to integrate ARC VPX 512-bit vector processing directly into the Actus platform. This allows complex edge AI tasks, such as sensor fusion and predictive maintenance models, to run on the exact same chip that drives the motor. Instead of sending sensor data back to a central brain for processing—a journey that introduces fatal latency—the joint itself becomes intelligent. It can anticipate mechanical wear, adjust to load shifts, and process tactile feedback instantaneously. Industry analysts note that while established players still lead in high-end visual perception for robotics, RISC-V is rapidly capturing the control plane, where deterministic real-time loops are non-negotiable.

Building Trust Through Deterministic Safety and Security

As a society, our willingness to integrate autonomous robots into our daily lives hinges entirely on our trust in their safety and security. A robot operating in a crowded warehouse or an assisted-living facility is not merely a machine; it is a heavy, kinetic object capable of causing significant harm if its systems are compromised or if it fails to react predictably to its environment.

MIPS has addressed these societal imperatives by weaving functional safety and advanced cybersecurity directly into the silicon of the Actus platform. The system integrates Gigabit Ethernet with three-port Time-Sensitive Networking (TSN), alongside CAN-FD and serial interfaces. This infrastructure ensures that joint controllers can exchange position and torque data with the central controller over a deterministic network, guaranteeing that critical safety commands are never delayed by lower-priority data traffic.

More notably, Actus includes a hardware security module fortified with post-quantum cryptography (PQC). As quantum computing advances, the encryption protocols that currently secure our digital and physical infrastructure are increasingly at risk. By embedding PQC at the node level, MIPS is future-proofing the robotic workforce against next-generation cyberattacks, ensuring that malicious actors cannot hijack the physical movements of autonomous machines.

Coupled with support for functional safety standards—targeting ISO 26262 and achieving levels up to ASIL-D and SIL3—the platform provides the rigorous, certified reliability demanded by regulatory bodies and liability-conscious manufacturers. Safety is no longer a software afterthought; it is hardcoded into the actuator's reflexes.

From Prototype to Production

The true test of any technological breakthrough is its commercial viability. For the robotics industry, transitioning from impressive laboratory prototypes to scalable, mass-manufactured products has been a notoriously difficult chasm to cross. The integration provided by GF’s FDX technology, complete with AutoPro150 qualification and embedded on-chip MRAM, significantly reduces board area and overall system cost, making commercial scaling economically feasible.

Recognizing that hardware is only as capable as the software that commands it, MIPS has adopted a "software-first" development approach. Engineering teams can begin writing and testing code on the MIPS Atlas Explorer virtual platform long before physical silicon arrives. This digital twin environment allows developers to simulate real-world workloads, optimize micro-architectural efficiency, and drastically accelerate the time-to-market. Once ready, teams can transition to the Actus-1 evaluation board and SDK for working prototypes, supported by an expanding ecosystem of partners, including Green Hills Software, to extend the software stack for safety-certified products.

As we stand on the precipice of a new era in automation, the focus is rightfully shifting from what AI can learn to what it can physically achieve. Platforms like MIPS Actus remind us that the future of robotics will not be built solely on algorithmic brilliance, but on the quiet, deterministic reliability of the silicon that translates those algorithms into safe, meaningful action in the physical world.