Huawei Details Open-Source AI Development Roadmap at Huawei Connect 2025
Estimated Reading Time: 6-7 minutes
- Huawei is committed to open-sourcing its entire AI software stack, including CANN, Mind series, and openPangu models, by December 31, 2025.
- The company candidly acknowledged past developer friction with Ascend infrastructure, indicating a renewed focus on building a developer-friendly ecosystem.
- CANN will feature open interfaces for its compiler and virtual instruction set, while the Mind series application enablement kits and toolchains will be fully open-source.
- Huawei plans extensive system integration, making the UB OS Component open-source for integration into existing OSes and prioritizing compatibility with popular AI frameworks like PyTorch and vLLM.
- Developers and organizations are urged to evaluate the upcoming releases meticulously and actively engage with the emerging community to shape the platform’s long-term viability.
- Key Takeaways
- Acknowledging Past Challenges and Paving the Way Forward
- The Pillars of Huawei’s Open-Source Commitment: CANN, Mind Series, and OpenPangu
- Beyond Core Software: System Integration and Framework Compatibility
- Navigating the Open-Source Future: Deadlines, Evaluation, and Unanswered Questions
- Frequently Asked Questions (FAQ)
Open-source AI development took centre stage at Huawei Connect 2025 last week, with Huawei laying out implementation timelines and the technical specifics around making its entire AI software stack publicly available by year-end. The announcements came with context that matters to developers: frank acknowledgement of past friction, specific commitments about what components will be released, and details about how the software will integrate with existing workflows and operating systems.
Acknowledging Past Challenges and Paving the Way Forward
Huawei’s Deputy Chairman and Rotating Chairman, Eric Xu, opened his keynote with unusual candour regarding the challenges developers have previously faced with Ascend infrastructure. This transparency signals a renewed commitment to fostering a developer-friendly ecosystem.
Referencing the impact of DeepSeek-R1’s release earlier this year, Xu noted: “Between January and April 30, our AI R&D teams worked closely to make sure that the inference capabilities of our Ascend 910B and 910C chips can keep up with customer needs.” This proactive response highlights Huawei’s dedication to performance and relevance in the rapidly evolving AI landscape.
Following extensive customer feedback sessions, Xu stated: “Our customers have raised many issues and expectations they’ve had with Ascend. And they keep giving us great suggestions.” This direct acknowledgement of developer pain points provided crucial context for the comprehensive open-source commitments announced at the August 5, 2025 Ascend Computing Industry Development Summit and reinforced by Xu at Huawei Connect.
For developers who have struggled with Ascend tooling, documentation, or ecosystem maturity, the frank assessment signals awareness of gaps between the platform’s technical capabilities and its practical usability. The open-source strategy appears designed directly to address these friction points by enabling community contributions, transparency, and external improvements. For instance, imagine a developer trying to debug a performance bottleneck in their AI model; with proprietary tools, they might hit a wall. An open-source approach would allow them to inspect the underlying compiler or runtime, potentially even contributing fixes or improvements that benefit the entire community.
The Pillars of Huawei’s Open-Source Commitment: CANN, Mind Series, and OpenPangu
Huawei’s open-source strategy is multifaceted, targeting different layers of its AI software stack with specific commitments and timelines.
CANN: Foundational Toolkit Open Interfaces
The most technically significant commitment involves CANN (Compute Architecture for Neural Networks), Huawei’s foundational toolkit that sits between AI frameworks and Ascend hardware. At the August summit, Xu specified: “For CANN, we will open interfaces for the compiler and virtual instruction set, and fully open-source other software.”
This tiered approach means developers can understand and potentially optimise how their code gets compiled for Ascend processors, even if the compiler implementation itself remains partially closed. The distinction matters for performance tuning, offering visibility for optimisation while retaining some proprietary elements. The timeline remains firm: “We will go open source and open access with CANN (based on existing Ascend 910B/910C design) by December 31, 2025.” This clarifies the release will reflect current-generation hardware.
Mind Series: Fully Open-Source Application Enablement
Beyond the foundational CANN layer, Huawei committed to open-sourcing what developers interact with daily: “For our Mind series application enablement kits and toolchains, we will go fully open-source by December 31, 2025,” Xu said at Huawei Connect, reinforcing the commitment made at the Ascend Computing Industry Development Summit on August 5, 2025.
The Mind series encompasses the practical development environment – SDKs, libraries, debugging tools, profilers, and utilities. This blanket commitment to full open-source means the entire application layer toolchain becomes inspectable, modifiable, and community-extensible. While the announcement didn’t specify which tools comprise the Mind series or their language support, the implications for community-driven development are substantial.
OpenPangu Foundation Models: Entering the Open Model Space
Huawei has also committed to “fully open-source our openPangu foundation models.” This strategically positions Huawei in the open-source foundation model space alongside major players like Meta’s Llama series and Mistral AI.
While specific details on capabilities, parameter counts, training data, or licensing terms were not provided, this move offers developers starting points for domain-specific applications without the need for massive computational resources for training from scratch. The December release will be crucial in determining whether openPangu models can stand as competitive alternatives.
Beyond Core Software: System Integration and Framework Compatibility
Addressing practical deployment challenges, Huawei also revealed its plans for broader ecosystem integration.
Operating System Integration Flexibility
A common barrier to adopting new AI infrastructure is operating system compatibility. Huawei announced that “Huawei has made the entire UB OS Component open-source, so that its code can be integrated into upstream open-source OS communities like openEuler.”
The integration approach offers unusual flexibility: “Users can integrate part or all of the UB OS Component’s source code into their existing OSes, to support independent iteration and version maintenance. Users can also embed the entire component into their existing OSes as a plug-in to ensure it can evolve in-step with open-source communities.” This modular design means organisations running Ubuntu, Red Hat Enterprise Linux, or other distros aren’t forced to migrate, significantly lowering deployment friction. However, this flexibility also shifts responsibility for testing, maintenance, and updates to the integrating organisations.
Framework Compatibility Strategy
Perhaps the most important factor for developer adoption is compatibility with existing AI frameworks. Huawei “has been prioritising support for open-source communities like PyTorch and vLLM to help developers independently innovate.” PyTorch compatibility is particularly significant given its dominance in AI research and production.
If developers can write standard PyTorch code that executes efficiently on Ascend hardware with minimal modifications, the barrier to experimentation drops substantially. Native vLLM support also targets the high-demand use case of optimised large language model inference. The quality and completeness of these integrations will ultimately determine their success in genuinely lowering adoption barriers.
Navigating the Open-Source Future: Deadlines, Evaluation, and Unanswered Questions
The December 31, 2025 timeline for open-sourcing CANN, Mind series, and openPangu models is just three months away, suggesting significant preparatory work is already complete.
Initial release quality will largely determine community response. Open-source projects thrive on comprehensive documentation, clear examples, and mature tooling. Successful open-source projects require sustained investment beyond initial code publication, encompassing community management, issue triage, pull request review, and documentation maintenance. Huawei’s long-term commitment to these aspects will define whether the platform develops an active contributor base.
What Remains Unspecified
Despite the specific commitments, important details remain undefined. Huawei hasn’t specified the licence selection for the December releases. Permissive licences (Apache 2.0, MIT) offer broad commercial use, while copyleft licences (GPL) require derivative works to also be open-sourced. Overall governance structures are also unclear: Will there be an independent foundation? Will Huawei accept external maintainers? How will roadmaps be decided? These governance questions are crucial for attracting genuine external participation.
Actionable Steps for Developers and Organizations:
- Prepare for Release Evaluation: Developers and organisations should spend the next three months assessing their technical requirements and evaluating whether Ascend hardware specifications align with their workload needs. Prepare teams for hands-on evaluation of the actual code, documentation, and tools once the December 31st release arrives.
- Plan for Flexible Integration: For organisations considering the UB OS Component, begin planning how its source code will integrate with existing operating systems. Identify internal Linux expertise or external resources needed for testing, maintenance, and ongoing updates, understanding that this component comes as open-source, not a fully supported product for arbitrary distributions.
- Engage with the Emerging Community: Post-December, actively participate in community forums, test the released components, file issues, and explore potential contributions. Your feedback and engagement will be crucial in shaping the platform’s evolution and indicating its long-term viability.
For developers and organisations considering investment, the next three months provide time for preparation. The December 31 release will provide concrete materials for hands-on evaluation. By mid-2026, patterns should emerge about whether Huawei’s open-source AI development strategy is succeeding in building an active community around Ascend infrastructure. A six-month window from December 2025 through to around mid-2026 will be an evaluation period for determining whether this open-source platform warrants serious investment of time and resources.
See also: Inside Huawei’s plan to make thousands of AI chips think like one computer
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and co-located with other leading technology events. Click here for more information.
AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.
Frequently Asked Questions (FAQ)
1. What is Huawei’s core commitment regarding AI open source?
Huawei has committed to open-sourcing its entire AI software stack, including foundational toolkits, application enablement kits, and foundation models, by December 31, 2025. This aims to foster a more transparent and developer-friendly ecosystem around its Ascend infrastructure.
2. What specific components of Huawei’s AI stack are being open-sourced?
The commitment includes CANN (Compute Architecture for Neural Networks) with open interfaces for compilers and virtual instruction sets, the Mind series application enablement kits and toolchains which will be fully open-source, and openPangu foundation models.
3. When can developers expect these open-source releases?
All committed open-source components, including CANN, Mind series, and openPangu models, are scheduled for release by December 31, 2025.
4. How is Huawei addressing past developer feedback?
Huawei’s Deputy Chairman, Eric Xu, candidly acknowledged past challenges and developer friction with Ascend infrastructure. The new open-source strategy is a direct response to this feedback, aiming to improve usability, enable community contributions, and increase transparency.
5. What operating systems and AI frameworks will be compatible with Huawei’s open-source AI?
Huawei is making its UB OS Component open-source for integration into upstream OS communities like openEuler, offering flexibility for users to integrate it into various existing operating systems. The company is also prioritizing support for popular AI frameworks like PyTorch and vLLM to ensure broader compatibility and ease of adoption.
