RISC-V at Foundry Scale: How MIPS and GlobalFoundries Industrialize Physical AI

As AI expands beyond cloud data centers into the machines that move, build, and protect the physical world, the silicon foundations beneath it must evolve. Physical AI—spanning automotive, industrial, aerospace, and networking systems—depends on deterministic performance and verifiable design. Traditional open-ISA approaches have struggled to deliver that rigor at industrial scale.

This HyperFRAME Research white paper, sponsored by MIPS, explores how RISC-V at Foundry Scale unites open-standard freedom with foundry-grade execution. By embedding MIPS’s proven multi-threaded architectures directly inside the GlobalFoundries design and manufacturing workflow, the partnership transforms open silicon from concept into repeatable, certifiable infrastructure for the physical world.

The paper introduces the STAC Framework—Sense, Think, Act, Communicate—which defines how Physical AI systems close the loop between perception, reasoning, control, and coordination. Supported by Atlas Explorer, MIPS’s digital-twin environment, and GlobalFoundries’ monthly shuttle cadence, this model delivers rapid prototype-to-production cycles and deterministic silicon performance verified at transistor depth.