top of page
Writer's pictureRich Washburn

The Chip Wars: Huawei’s Ascend 910C and Ampere’s Rise in AI Computing


Audio cover
The Chip Wars

The world of AI chip development is a rapidly evolving battlefield, with key players vying for dominance in a market that could shape the future of technology. Two stories have recently emerged that highlight the fierce competition and innovation driving this industry: Huawei's bold challenge to Nvidia with its Ascend 910C chip and Ampere Computing’s quiet disruption in the cloud computing sector. These developments are not just about technological advancements; they represent strategic moves in a global contest for supremacy in AI and computing power.


Huawei’s Ascend 910C: A Strategic Counter to Nvidia’s Dominance


Huawei, a company that has faced intense scrutiny and sanctions from the United States, is making waves with the introduction of its Ascend 910C chip. This chip is positioned as a direct competitor to Nvidia's H100 GPU, which has been the gold standard for AI computing. For years, Nvidia has held a dominant position in the AI chip market, especially in China, where it once controlled over 90% of the market share. However, U.S. restrictions have significantly impacted Nvidia’s ability to sell its most advanced chips in China, opening the door for competitors like Huawei.


The Ascend 910C is more than just a chip; it is a statement of Huawei’s resilience and its ambition to lead in the AI sector. Despite facing significant challenges, including U.S. sanctions that have made it difficult to access advanced manufacturing tools, Huawei has managed to develop a chip that rivals Nvidia's best. Early tests by major Chinese tech companies like ByteDance and China Mobile have shown that the Ascend 910C can perform on par with the H100, especially in areas crucial for AI development, such as training large models and processing massive datasets.


However, Huawei’s journey is not without its hurdles. Reports suggest that the company is struggling with manufacturing yields, with up to 80% of the chips being defective. This issue is largely due to the limitations of its manufacturing partner, SMIC, which lacks the advanced EUV technology required for producing 7nm chips. Instead, SMIC has been forced to rely on less advanced DUV machines, using complex multi-patterning techniques that have proven difficult to execute with high yields. Huawei is now in the process of redesigning the Ascend 910C to improve yields, a move reminiscent of its previous efforts with the Kirin chip.


Ampere Computing: A Disruptive Force in Cloud AI


While Huawei battles to overcome manufacturing challenges, another company, Ampere Computing, is quietly disrupting the cloud computing industry. Founded by former Intel president Renee James, Ampere has focused on developing chips based on ARM architecture, targeting the cloud market with its high-performance, energy-efficient processors.


Ampere’s latest chip, the AmpereOne, is a 256-core powerhouse designed specifically for cloud-native applications. Unlike traditional CPUs that focus on increasing core size, Ampere’s strategy is to maximize the number of cores, providing better performance per watt—an essential factor in the energy-intensive world of data centers. This approach has resonated with major cloud providers like Google, Amazon, and Oracle, who have adopted Ampere’s chips for their operations.


However, Ampere faces a significant challenge as its early customers, including Amazon and Google, begin to develop their own chips. This trend toward vertical integration—where companies control both the hardware and software stacks—could threaten Ampere’s business model. To stay competitive, Ampere’s chips will need to offer superior performance or cost advantages, a difficult task in the increasingly competitive AI chip market.


The stories of Huawei and Ampere highlight the intense competition and rapid innovation driving the AI chip industry. For Huawei, the ability to overcome manufacturing difficulties and deliver a reliable product will determine whether it can truly challenge Nvidia's dominance. For Ampere, the key to success lies in maintaining its technological edge as its customers move toward self-reliance.


As AI continues to integrate into every aspect of modern life, from cloud computing to consumer electronics, the stakes in this competition have never been higher. The companies that succeed in developing the most advanced and efficient AI chips will not only dominate the tech industry but also shape the future of global power and security.


For the broader industry and consumers, these developments could lead to faster, more powerful, and more affordable AI solutions. As companies like Huawei and Ampere push the boundaries of what’s possible, we are on the brink of a new era in technology—one where the smallest chips could have the most significant impact.


The race is on, and the outcome could reshape the world as we know it. What are your thoughts on these developments? Join the conversation and stay tuned for more insights into the fast-paced world of AI innovation.


Comments


bottom of page