IBM and Arm Partner to Run AI Software on Mainframes
April 6, 2026 - 12:57 pm
In short:
IBM and Arm announced a strategic collaboration on April 2, 2026, to enable Arm-based software to run on IBM Z and LinuxONE mainframes—the platforms that process most of the world’s regulated enterprise transactions. The partnership focuses on three areas: virtualisation to host Arm software environments on IBM hardware, security and compliance for regulated industries, and long-term ecosystem interoperability. The goal is to bring Arm-native AI software closer to enterprise data that IBM Z customers cannot move to the public cloud. No shipping date was provided by IBM.
The Problem the Partnership Addresses:
IBM Z and LinuxONE mainframes run on IBM’s s390x architecture, while the modern AI software ecosystem, including PyTorch, TensorFlow, llama.cpp, ONNX Runtime, and Kubernetes workloads, is primarily developed for x86 and Arm architectures. Porting this software to s390x is time-consuming, expensive, and often lags behind upstream development.
For enterprises relying on IBM Z for critical transactions, data storage, and compliance-sensitive workloads, this creates a significant gap between where their data resides and where AI inference happens—often in the cloud on different hardware.
Partnership Focus Areas:
The collaboration is structured around three main workstreams:
- Virtualisation: Building solutions to host Arm software environments on IBM hardware.
- Security and Compliance: Ensuring that Arm-based software meets the stringent security and compliance requirements of regulated industries.
- Long-term Ecosystem Interoperability: Fostering a shared ecosystem where AI tooling runs seamlessly on both Arm and s390x architectures.
Caveat:
While the partnership aims to bridge the gap, it's important to note that IBM has not specified a timeline for implementing these changes.