Huawei launches Atlas 960E SuperPoD for large-scale AI computing
Global technology brand Huawei has launched the Atlas 960E SuperPoD, a new AI computing solution designed to support growing demand for large-scale artificial intelligence applications.
Huawei unveiled the solution at HUAWEI CONNECT 2026 in Shanghai, where Deputy Chairman of the Board and Rotating Chairman David Wang introduced it during his keynote address on advancing the intelligent world, according to the press release.
Held under the theme "Advancing the Agentic World", the event brought together industry leaders, customers, partners and technology experts to discuss the role of AI, digital connectivity and collaboration in creating a more intelligent and productive future.
Keep updated, follow The Business Standard's Google news channel
The Atlas 960E SuperPoD is designed to support the development, training and operation of advanced AI models.
Huawei said the system combines large-scale computing performance, energy efficiency and reliable operations to help organisations build and manage AI infrastructure more effectively.
The company said the Atlas 960E is its first SuperPoD based on near-packaged optics technology and is designed to support the training and inference requirements of increasingly large AI models.
Huawei also highlighted its Atlas 950 SuperPoD and SuperPoD-based SuperClusters, its first-generation products built on its Peerium Computing Architecture.
The Peerium Computing Architecture is designed to enable processors at a scale of up to one million to work together as a single computing system through nested parallelism, unified memory addressing and peer-to-peer interconnection.
Huawei said the architecture is intended to meet increasing demand for computing power as AI models and applications continue to expand in scale and complexity.
The company also unveiled an upgraded TaiShan 950 SuperPoD for general-purpose computing and OceanStor M900, a storage solution designed to support AI-related data and context-memory requirements.
The TaiShan 950 applies the SuperPoD architecture to general-purpose computing, while OceanStor M900 is designed to provide large-scale storage support for AI workloads.
Together, the solutions are intended to help industries manage increasing computing and data demands and accelerate the practical adoption of AI.
Huawei said its UnifiedBus technology provides the interconnection foundation for its SuperPoDs and SuperClusters, allowing computing, storage and networking resources to work together more efficiently.
The company said multiple SuperPoDs can connect to form larger SuperClusters capable of supporting computing systems with up to one million neural processing units.
Huawei also said its CANN platform has moved into sustained, community-driven open-source development, allowing developers, partners, and the wider technology community to participate in developing the AI computing ecosystem.
The company said its open-source and open-system strategy is intended to make AI development tools and computing capabilities more accessible while encouraging broader collaboration across the technology ecosystem.
"At Huawei, we are committed to building powerful AI infrastructure and promoting open-source and open systems to strengthen the computing ecosystem," David Wang said.
"We will continue to work with partners to support AI innovation and advance the intelligent world," he added.
Huawei said it continues to work with customers and partners across industries, including telecommunications, finance, manufacturing, energy, transportation, public services and education, to support digital transformation.
The company said its AI infrastructure, computing architecture, and open ecosystem initiatives aim to help industries adopt intelligent technologies and build a more connected and intelligent future.
