A week after Huawei announced its decision to open source the CANN (Neural Network Computing Architecture) software toolkit, Tech Industry is still handling the implications of this movement for the future of AI development.
By making Huawei Cann Cuda’s open source available to developers around the world, the Chinese tech giant fired what many consider to be a key shot in Nvidia’s two years and the battle against the ongoing dominance of AI computing.
While that is a prominent challenge to the current situation, the real question is whether Huawei can overcome the substantial technical and systematic barriers that have maintained Cuda virtually unchallenged for nearly 20 years.
What is Cann and why is it important?
Cann is Huawei’s heterogeneous computing architecture, offering a multi-level programming interface that helps developers build AI applications optimized for Huawei’s Ascend AI GPUs. First introduced in 2018 as part of Huawei’s AI strategy, Cann will serve as the equivalent of Nvidia’s Cuda platform.
Cann provides APIs for ASCEND’s AI applications, providing developers with several options for building high-level and performance-intensive applications. This architecture represents the years of development aimed at creating a comprehensive software ecosystem centered around Huawei’s AI hardware.
Strategic Timing Behind Open Source Decisions
Huawei’s decision to create a Cann Open-Source will be a particularly tense moment in the US and China technology relationship. Huawei’s revolving chairman Eric Xu Zhijun said the move will help “speed up innovation from developers” and “make the rise easier” during the company’s developer meeting in Beijing.
The announcement follows closely after China’s Cyberspace Management (CAC) began an investigation into what is called “serious security issues,” including Nvidia’s processors, and based on requests from US lawmakers to add tracking capabilities from US lawmakers.
Regulation scrutiny adds another layer of complexity to the already tense relationship between the two superpowers.
CUDA’s exclusive grip on AI development
To understand the importance of Huawei’s movement, it is important to examine Nvidia’s Cuda control. Cuda is often referred to as a closed “moat,” or sometimes called a “Swamp,” and has been seen as a barrier for developers to seek cross-platform compatibility.
The tight integration with NVIDIA hardware has locked developers into a single vendor ecosystem over the last 20 years, with every effort to bring CUDA to other GPU architectures through the translation layer being blocked by the company. An additional provision has been added to the CUDA licensing agreement that prevents developers from running CUDA on third-party GPUs via the translation layer.
Many Chinese AI developers use NVIDIA GPUs for the CUDA platform, which has been the default development platform for many years. This situation highlights the challenges Huawei faces when convincing developers to move into that ecosystem.
Industry analysis and market impact
Technology analysts offer a mixed evaluation of Huawei’s open source strategy. Open-sourcing CAN helps Huawei accelerate the adoption of its in-house software toolkit and accelerate its hardware, but it could take years for Cann to match CUDA’s ecosystem support.
The competitive landscape reveals the magnitude of Huawei’s challenge. Even in open source status, adoption can depend on the extent to which it supports existing AI frameworks, particularly for large-scale language models and new workloads of AI writer tools. The software ecosystem around CUDA contains thousands of optimized libraries and extensive documentation that took years to develop.
However, there are some claims that Huawei’s hardware has shown signs of progress, with certain ascend chips being able to outperform Nvidia processors under certain conditions. Reports suggest that Huawei’s performance trajectory fills the performance gap for Nvidia running Deepseek R1 on CloudMatrix 384.
Building an alternative ecosystem
Huawei, according to South China Morning Postbegan discussions with major AI users in China, universities, research institutes and business partners. The collaborative approach reflects successful open source initiatives in other technology sectors where community contributions accelerate development and adoption.
Context of the Global Chip War
The open source Cann initiative is fitted with China’s technological independence. The country’s open source drive is gaining momentum, allowing more domestic tech companies to expose their own technologies. Recent examples include Xiaomi open sourcing of the leading Midashenglm-7B audio language model and the release of Alibaba, the QWEN3-CODER AI coding model.
All of this comes amidst the backdrop of continuing US export restrictions targeting Chinese technology companies. In today’s environment where US restrictions affect Huawei’s hardware exports, building a robust domestic software stack of AI tools is just as important as improving chip performance.
Expert skepticism and future challenges
Raw performance alone does not guarantee developer migration without comparable software stability and support. The challenge is to include integration into document quality, community activities and development workflows, beyond technical capabilities.
The road ahead
The impact on the global semiconductor industry remains important. As US-China technology contests intensify, Huawei’s open source strategy represents a shift towards building co-ecosystems that can reconstruct the way AI software development evolves globally.
It remains to be seen whether this initiative will succeed in Nvidia’s dominance, but it certainly marks a new chapter in the ongoing battle to control AI computing infrastructure that drives next-generation innovation.
reference: Alan Turing Institute: Humanities are the key to the future of AI.
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo in Amsterdam, California and London. The comprehensive event will be held in collaboration with other major events, including the Intelligent Automation Conference, Blockx, Digital Transformation Week, and Cyber Security & Cloud Expo.
Check out other upcoming Enterprise Technology events and webinars with TechForge here.
