Zyphra announced the launch of Zyphra Cloud on May 4 2026, a full‑stack AI platform that runs on AMD Instinct MI355X GPUs built on AMD’s CDNA architecture and featuring HBM3E memory.
The platform includes Zyphra Inference, a serverless inference service that supports long‑horizon agentic models such as DeepSeek V3.2, Kimi K2.6, and GLM 5.1, and is engineered to deliver high‑throughput, low‑latency performance for production‑grade inference workloads.
Zyphra partnered with TensorWave to provide the dense, liquid‑cooled cluster infrastructure that hosts the MI355X GPUs, making Zyphra Cloud one of the first cloud services to deploy this accelerator in a production environment. The launch expands AMD’s reach beyond its traditional hyperscaler customers and demonstrates real‑world performance of the MI355X in demanding open‑weight models.
The partnership signals confidence from an independent AI‑focused firm in AMD’s accelerator technology and could boost demand for future MI400 and MI450 series GPUs. Zyphra, founded in 2020 and a unicorn since October 2025, views this launch as a major operational milestone that positions it to capture a larger share of the AI infrastructure market.
The launch underscores AMD’s AI roadmap, with the MI355X as the current offering and the MI400 series slated for release in the second half of 2026, aligning Zyphra’s strategy with AMD’s goal to compete against dominant players and showcase the capabilities of its hardware in demanding open‑weight models.
The content on EveryTicker is for informational purposes only and should not be construed as financial or investment advice. We are not financial advisors. Consult with a qualified professional before making any investment decisions. Any actions you take based on information from this site are solely at your own risk.