Zyphra Cloud, a full-stack neocloud powered by AMD, announces 15 megawatts of MI355X GPU capacity.
SAN FRANCISCO, May 11, 2026 /PRNewswire/ — Zyphra today announced the availability of 15 megawatts of AMD Instinct™ MI355X GPU capacity through Zyphra Cloud, expanding the platform with software-driven infrastructure for bare-metal AMD deployments.
Zyphra Cloud is a full-stack neocloud powered by AMD, designed for AI-native startups, enterprises, and frontier AI hyperscalers. The platform launched with serverless inference for leading open-source models on MI355X GPUs and now expands to bare-metal AMD infrastructure, enabling customers to run advanced AI workloads across pretraining, large-scale reinforcement learning & post-training, and agentic inference.
“AMD has rapidly emerged as a major compute platform for AI, and our goal is to accelerate adoption by bringing the systems and software work developed through Zyphra Research directly into production infrastructure,” said Krithik Puthalath, Founder and CEO of Zyphra. “Zyphra Cloud is not just another AMD neocloud. We’ve demonstrated expertise across the full AI lifecycle, from pre-training to production inference, and now deliver that as software-driven infrastructure to enable customers to go-to-market faster on AMD.”
Zyphra Cloud offers two primary deployment models: on-demand bare-metal GPU clusters for flexible workloads, and custom hyperscale AMD infrastructure for large-scale training and inference deployments. These offerings serve a range of customers from AI-native startups to enterprises and frontier AI hyperscalers.
Zyphra has designed and operated custom AMD clusters with IBM using Instinct GPUs and Pensando Pollara networking to run large-scale training and inference workloads. Building on this foundation, Zyphra Research has trained models on full-stack AMD systems and developed next-generation inference algorithms for production environments.
Zyphra Cloud brings these capabilities together into a unified platform, combining AI software with custom AMD infrastructure to deliver production AI systems. With 15 megawatts of MI355 capacity now available, customers can deploy and scale AI workloads on AMD with speed and reliability.
Software-Driven Infrastructure
Over the last week, Zyphra Research showcased AMD in production across the full AI lifecycle through a series of releases spanning MoE pretraining, reinforcement learning & post-training, and inference.
This included the release of ZAYA1-8B and ZAYA1-74B-Preview, models pretrained end-to-end on AMD Instinct MI300X infrastructure with AMD Pensando Pollara networking. Zyphra Research also introduced TSP, a next-generation inference algorithm designed for scalable long-context and agentic inference. Zyphra Cloud also powers agentic inference for large-scale open models such as DeepSeek and Kimi on AMD Instinct MI355X GPUs through serverless endpoints.
These systems and software optimizations now directly power Zyphra Cloud and are delivered as a service, enabling customers to deploy and scale custom AMD infrastructure reliably for production AI workloads.
Availability
AMD Instinct MI355 GPU capacity is available today through Zyphra Cloud. Zyphra plans to expand support to next-generation AMD platforms, including the MI450 series and beyond, as part of its roadmap to scale software-driven infrastructure across future AMD architectures.
For deployment options, reserved capacity, and enterprise infrastructure, contact sales at zyphra.com/contact-sales.
About Zyphra
Zyphra is an open superintelligence research and product company based in San Francisco, CA, building full-stack AI systems across research, infrastructure, and applications. Zyphra’s mission is to build human-aligned AI that helps individuals and organizations reach their fullest potential.
For more information, visit www.zyphra.com.
Media Contact
Paul White
Chief Business Officer
[email protected]
www.zyphra.com
SOURCE Zyphra
