China’s AI hardware ambitions test Nvidia’s dominance amid global tech realignment
Huawei is making a calculated move to extend its artificial intelligence (AI) hardware footprint beyond China by introducing its Ascend 910B AI chips to buyers in the United Arab Emirates, Saudi Arabia, and Thailand. These early-stage export discussions represent Huawei’s broader ambition to challenge Nvidia’s global influence, particularly in emerging digital economies open to alternatives amid U.S. trade restrictions.
While the export negotiations are still in early stages, Huawei is simultaneously expanding domestic deployments of its CloudMatrix 384 system—a supercomputing setup powered by its more advanced Ascend 910C chips. Together, these efforts reflect a two-pronged strategy: promote international growth while strengthening its position at home.
The rise of Ascend and CloudMatrix
The Ascend 910B is Huawei’s earlier-generation AI chip designed to offer decent performance while remaining exportable under current U.S. sanctions. Though it trails Nvidia’s latest offerings, it delivers enough computational power for many generative AI tasks, particularly for governments and businesses in developing markets.
Meanwhile, CloudMatrix 384—Huawei’s flagship domestic AI infrastructure—has become a powerful showpiece of China’s AI hardware ambition. This system combines 384 Ascend 910C chips to deliver an estimated 300 petaflops of compute power. However, the CloudMatrix system isn’t ready for export yet, mainly due to supply limitations and regulatory complexities.
By offering Ascend 910B overseas while promoting CloudMatrix 384 inside China, Huawei signals its intent to maintain competitiveness on both global and domestic fronts.
Huawei’s international chip push
Huawei’s push into the Middle East and Southeast Asia focuses on selling low-thousand units of the Ascend 910B—a modest volume that allows Huawei to test foreign interest and local adoption. It also reflects a measured and risk-aware market entry strategy.
In the UAE, early conversations have taken place but haven’t yet materialized into firm orders. Saudi Arabia’s interest seems to be growing, likely driven by the country’s digital transformation push under its Vision 2030 program. In Thailand, where discussions remain fluid, the market potential lies in government-backed innovation zones and AI-driven industry adoption.
What unites these target markets is their growing desire for non-U.S. AI hardware alternatives, especially as export bans tighten and Western chips become harder to access. Huawei’s approach, although cautious, could find resonance in markets that want cost-effective, geopolitically neutral solutions.
CloudMatrix deployment and developer traction
Inside China, Huawei continues to scale its AI offerings by expanding remote access to the CloudMatrix 384 system. Data centers in Anhui and Inner Mongolia have already integrated the setup, allowing developers, research institutions, and enterprise clients to access supercomputing power through cloud-based usage models.
Huawei is also preparing to launch its next-generation Ascend 910D and Ascend 920 chips by late 2025. These new models are expected to offer significant performance improvements and will likely be positioned as domestic competitors to Nvidia’s H100 series—especially in fields like large language models (LLMs), AI agents, and robotics.
This dual-track strategy—exporting older chips abroad while advancing infrastructure and R&D at home—helps Huawei manage geopolitical risk while building technological credibility.
China’s AI hardware blueprint takes shape
Huawei’s chip strategy offers a window into China’s broader push for tech sovereignty. While CEO Ren Zhengfei has acknowledged that Huawei’s chips lag behind Nvidia’s, the company is focusing on vertical integration—combining software, optics, and system design to narrow the gap.
China’s long-term vision prioritizes self-reliance, especially in high-stakes sectors like semiconductors and AI computing. Huawei’s CloudMatrix rollout aligns with government policies that encourage import substitution and domestic innovation.
At the same time, the attempt to gain a foothold in foreign markets—especially in politically aligned or neutral countries—signals that Huawei sees AI chips not just as products, but as strategic tools in reshaping global technology alliances.
What success could look like for Huawei
Huawei’s long-term position in the global AI chip market will depend on several variables:
Securing export deals in at least one major Gulf or Southeast Asian market will validate Huawei’s outreach strategy.
Successful deployment of CloudMatrix systems will demonstrate that performance limitations can be mitigated through system-level innovation.
Ecosystem development—such as building software libraries, developer communities, and partner integrations—will be critical for lasting impact.
If Huawei can prove that Ascend chips offer value, even without the raw power of Nvidia’s GPUs, it may find itself at the forefront of AI hardware diversification. Furthermore, as global demand for AI infrastructure surges, cost-effective and geopolitically neutral solutions could become attractive alternatives, especially in the Global South.
Huawei enters the global AI chip race
Huawei’s efforts to market Ascend 910B chips to Middle Eastern and Southeast Asian countries marks a significant milestone in China’s evolving AI ambitions. While still early, this move highlights how global tech competition is shifting—not just through software and algorithms but also through hardware and infrastructure.
With CloudMatrix 384 expanding at home and export strategies taking shape abroad, Huawei could emerge as a credible alternative to Nvidia. Whether that transformation happens in months or years will depend on performance parity, ecosystem maturity, and political will.
But one thing is certain: Huawei’s AI hardware campaign is no longer just about catching up—it’s about changing the terms of the competition.
