Huawei Targets AI Innovation with New Ascend 910D Chip

China's race for technological independence gains momentum as Huawei develops a new AI processor designed to challenge Nvidia’s dominance.

As countries adapt to shifting global supply chains and export controls, China's access to advanced technology narrows.

Huawei responds with the Ascend 910D, aiming to match the power of US-restricted hardware.

Huawei pursues AI innovation with Ascend 910D

The launch of Huawei’s Ascend 910D marks a bold move within the AI technology sector.

According to Reuters, Huawei hopes this processor will deliver performance comparable to Nvidia’s H100 graphics processing unit (GPU), currently a leader in AI model training.

The Ascend 910D reflects China's strategic push to develop indigenous semiconductor capabilities, especially as Washington tightens its grip on advanced technology exports.

Huawei's effort builds upon its earlier work with the 910C chip, expected to begin mass shipments to Chinese clients as early as next month.

This reveals a multi-generation roadmap, as the company targets multiple layers of the AI computing market with varied solutions.

Huawei and other Chinese chipmakers have faced ongoing challenges in matching Nvidia’s technical achievements.

Nvidia’s specialised processors, paired with an entrenched ecosystem, have made it an industry standard for AI development.

In particular, Nvidia’s GPUs excel in model training, a process where vast quantities of data refine algorithm decision-making over time.

The tightening of US export rules has only sharpened these challenges.

Restrictions began when the US government barred the sale of Nvidia’s H100 chip to Chinese customers in 2022, even before the chip’s official launch.

Controls now include Nvidia’s newer B200 chip, leaving Chinese tech companies struggling to access the hardware essential for large-scale AI model training.

Export controls and China’s push for semiconductor innovation

US export controls reshape the landscape for China’s AI researchers and technology companies.

Training state-of-the-art AI models demands immense computing power and restrictions have made acquiring leading-edge semiconductors far more difficult.

Huawei’s new chip is part of the answer, but the gap with Nvidia goes beyond hardware.

A crucial difference lies in the software ecosystem.

Nvidia’s Compute Unified Device Architecture (CUDA) platform has become a global standard, giving developers the tools they need to use GPUs for more than just graphics, enabling breakthroughs in AI, scientific computing and more.

In contrast, Huawei’s Compute Architecture for Neural Networks (CANN) platform still lags in adoption and developer support.

While the Ascend 910D shows technological promise, Huawei faces a steep road to broader acceptance.

Without the mature tools, libraries and community built around CUDA, developers may hesitate to switch platforms.

This software gap could limit Huawei’s chips to domestic markets, where the Chinese government support encourages adoption.

However, Huawei’s ambition is undiminished.

The company has pivoted aggressively into enterprise technology, cloud computing and semiconductor innovation since it was placed on the US Entity List in 2019, restricting its access to American components and software.

The Ascend 910D represents a crucial step in Huawei’s broader goal of maintaining relevance and leadership in the high-growth AI sector.

Huawei’s broader strategy for AI and technological independence

Huawei’s expansion into AI chip development fits into a wider national and corporate strategy.

China’s government has prioritised semiconductor self-sufficiency, pouring billions of dollars into research, development and manufacturing support for domestic firms.

Huawei stands at the centre of this effort, focusing on creating alternatives to American technology.

Achieving true technological independence requires more than matching hardware.

It demands building entire ecosystems of hardware, software and developer communities.

Huawei’s CANN platform aims to rival CUDA, but global adoption remains limited.

Outside of China, developers overwhelmingly build for Nvidia’s GPUs due to the reliability, scalability and compatibility of its platform.

Nevertheless, Huawei’s progress is notable.

Overcoming hardware limitations imposed by export controls shows resilience and innovation.

If Huawei can close the software gap and build a global developer base, the Ascend 910D could represent more than just an incremental product release.

It could symbolise a new chapter in China’s pursuit of technological leadership in AI.

Huawei's ongoing innovation also reflects the wider impact of geopolitical tensions on the global technology sector.

As the world fractures into competing technological spheres, companies like Huawei are working to build alternatives to Western-dominated ecosystems.

Although Huawei’s Ascend 910D faces strong competition from Nvidia’s established products, it signals China’s determination to lead in AI innovation despite external pressures.

Explore the latest edition of Technology Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.

Discover all our upcoming events and secure your tickets today.

Technology Magazine is a BizClik brand