Ant Group, the financial technology giant backed by Jack Ma, has announced a major artificial intelligence breakthrough, successfully training large-scale models using domestically produced Chinese semiconductors. This achievement marks a significant step in reducing dependency on foreign chips, particularly those from Nvidia, while cutting AI training costs by approximately 20%.
A New Era in AI Development
Sources familiar with the matter revealed that Ant utilized chips from Chinese manufacturers, including its affiliate Alibaba Group Holding Ltd. and Huawei Technologies Co., to train models using the Mixture of Experts (MoE) machine learning technique. These models reportedly performed at levels comparable to those trained on Nvidia’s H800 GPUs, which have been restricted from export to China due to U.S. sanctions.
Despite continuing to rely on Nvidia for some AI processes, Ant is increasingly turning to alternatives, including chips from Advanced Micro Devices Inc. and other Chinese suppliers, as it refines its AI capabilities. This shift underscores the growing effort by Chinese tech firms to become self-sufficient in AI infrastructure, circumventing Western-imposed technology restrictions.
China’s AI Race Accelerates
Ant’s latest AI initiative positions it within the intensifying competition between China and the U.S. in artificial intelligence innovation. The progress made by Ant follows the recent success of DeepSeek, a Chinese AI research firm that demonstrated how cost-effective AI models can rival those of OpenAI and Google DeepMind. Ant’s development strategy echoes this trend, proving that China is rapidly advancing in AI despite geopolitical tensions and restrictions on high-end hardware imports.
A research paper published by Ant Group this month asserts that its models, in some instances, outperformed Meta’s AI benchmarks. Although Bloomberg News has not independently verified these claims, the potential impact of such a breakthrough is undeniable. If proven accurate, Ant’s AI models could pave the way for more cost-efficient AI services in sectors such as finance, healthcare, and digital commerce.
The Cost-Efficiency Factor
One of the major challenges in AI development is the high cost of training large-scale models. Nvidia’s high-end GPUs, essential for complex computations, come at a steep price, limiting accessibility for smaller firms and research institutions. Ant has been actively working on strategies to mitigate these costs. According to its research, training a model with 1 trillion tokens using high-performance hardware currently costs approximately 6.35 million yuan ($880,000). By leveraging Chinese chips and optimized techniques, Ant claims it can reduce this cost to 5.1 million yuan—a significant improvement in efficiency.
Strategic Implications for AI’s Future
Ant’s AI progress comes at a time when Nvidia’s Chief Executive Officer, Jensen Huang, has been advocating for ever more powerful GPUs to meet the surging computational demands of AI. However, companies like Ant and DeepSeek are demonstrating that alternative approaches—such as MoE—can achieve comparable results without relying on the most expensive, high-performance chips.
Ant’s adoption of this strategy suggests a fundamental shift in the AI landscape, where efficiency and cost reduction become as critical as raw processing power. This could lead to a new wave of AI adoption, particularly among companies looking to deploy AI solutions without exorbitant infrastructure investments.
AI in Healthcare and Finance
Beyond cost-saving benefits, Ant is leveraging its AI models for real-world applications. The company recently acquired Chinese medical platform Haodf.com, expanding its AI-driven healthcare services. On Monday, Ant announced that its AI models are now deployed in seven hospitals across China, providing medical consultancy and diagnostics. The models integrate insights from DeepSeek R1, Alibaba’s Qwen, and Ant’s proprietary LLM, supporting healthcare professionals in streamlining patient care.
Additionally, Ant has launched two AI medical agents—Angel, which has assisted over 1,000 medical facilities, and Yibaoer, which specializes in medical insurance services. Last year, the company introduced the AI Healthcare Manager feature within Alipay, further demonstrating its commitment to integrating AI into critical industries.
China’s AI Self-Sufficiency on the Rise
As Ant Group pushes forward with its AI advancements, industry analysts note that China’s technology sector is making substantial progress toward self-reliance. Bloomberg Intelligence analyst Robert Lea highlighted that Ant’s research represents a pivotal moment in China’s AI evolution, as companies develop efficient AI models independent of Western technology.
While global AI leaders like OpenAI and Google continue to refine their multimillion-dollar AI architectures, China’s shift toward cost-effective alternatives could democratize AI development and deployment. As the competition intensifies, companies worldwide may be forced to reconsider their reliance on premium GPUs in favor of more scalable, cost-efficient solutions.
For more details on Ant Group’s AI initiatives, read the full report on Bloomberg: Jack Ma-Backed Ant Touts AI Breakthrough Using Chinese Chips.