Huawei's open-source gambit: Chinese tech giant makes SuperPoD architecture freely available

Company releases UnifiedBus 2.0 specifications as it positions for long-term competition in AI infrastructure market.

Huawei is taking an unexpected approach to competing in the global AI infrastructure race: giving away its core technology. At HUAWEI CONNECT 2025 in Shanghai last week, the Chinese tech giant announced it would make its SuperPoD architecture and UnifiedBus interconnect protocol freely available to industry partners, marking a significant shift in strategy as the company navigates ongoing semiconductor restrictions.

The move centers around Huawei's UnifiedBus 2.0 protocol, which enables thousands of AI processing units to work together as a single logical machine. Yang Chaobin, Huawei's Director of the Board and CEO of the ICT Business Group, said the company aims to collaborate with more industry players to create a solid computing backbone for all scenarios using this new architecture.

"We are committed to our open-hardware and open-source-software approach that will help more partners develop their own industry-scenario-based SuperPoD solutions. This will accelerate developer innovation and foster a thriving ecosystem," Yang explained during his keynote.

The technical specifications release comes as Huawei unveiled two flagship products: the Atlas 950 SuperPoD with up to 8,192 Ascend NPUs and the Atlas 960 SuperPoD with 15,488 NPUs. According to company executives, these systems represent the world's most powerful SuperPoDs based on publicly announced competitor roadmaps.

But the open-source strategy appears driven by practical constraints rather than pure altruism. Eric Xu, Huawei's Rotating Chairman, reiterated that the company's goal is to sustainably meet long-term computing demand by building SuperPoDs and SuperClusters with the semiconductor manufacturing process nodes that are practically available to the Chinese mainland.

This acknowledgment of semiconductor limitations provides context for Huawei's openness approach. By making its interconnect technology freely available, the company is essentially betting that ecosystem growth will compensate for hardware disadvantages stemming from US trade restrictions that limit access to advanced chip manufacturing processes.

The strategy extends beyond hardware architecture. On August 5, 2025, Huawei held the Ascend Computing Industry Development Summit in Beijing, where Xu shared four major commitments including fully open sourcing CANN compiler tools, Mind series application kits, and openPangu foundation models by December 31, 2025.

Industry deployment suggests real market traction despite geopolitical headwinds. Over 300 Atlas 900 A3 SuperPoD units have already been shipped in 2025, deployed for more than 20 customers from multiple sectors, including Internet, finance, carrier, electricity, and manufacturing sectors.

The technical capabilities Huawei claims are substantial. The Atlas 950 SuperPoD will deliver industry-leading performance across multiple key metrics, including the number of NPUs, total computing power, memory capacity, and interconnect bandwidth. The company projects its systems will maintain performance leadership "for years to come" based on competitor roadmaps.

However, the open-source pivot raises questions about Huawei's long-term monetization strategy in AI infrastructure. Xu explicitly stated that "our monetization strategy for AI is focused on hardware," suggesting the company plans to profit from chip sales rather than software licensing, even as it gives away previously proprietary technologies.

The approach contrasts sharply with competitors like NVIDIA, which maintains tight integration between hardware and software stacks. Whether Huawei's bet on ecosystem openness can overcome hardware constraints while generating sustainable revenue remains to be tested in increasingly competitive global AI infrastructure markets.

For channel partners and system integrators, the immediate opportunity lies in accessing proven large-scale AI infrastructure designs without licensing fees, potentially accelerating deployment timelines for enterprise AI projects across sectors where Huawei maintains market presence.