- Ant Group Co., backed by Jack Ma and affiliated with Alibaba (BABA), has utilized Chinese-made chips from Alibaba and Huawei, alongside U.S. chips from AMD, to train AI models with a 20% cost reduction, achieving performance similar to Nvidia’s (NVDA) chips, as reported by Bloomberg.
- The company’s Mixture of Experts (MoE) approach, detailed in a recent paper, leverages lower-cost hardware to cut AI training time and costs, reducing reliance on Nvidia and aligning with industry trends amid U.S. restrictions on advanced semiconductor exports to China.
- Ant announced upgrades to its AI healthcare solutions on Monday, built on DeepSeek’s R1 and V3, Alibaba’s Qwen, and its BaiLing model, now serving seven major hospitals in China, while Nvidia’s stock rose 2.06% to $120.13 despite the competitive news.
Ant Group Co., the Alibaba (BABA)-affiliate powerhouse behind Alipay – one of China’s two dominant mobile payment platforms – has been making notable advancements in artificial intelligence by utilizing a blend of Chinese and U.S.-made semiconductors to enhance its AI model training. According to a Bloomberg report, the Jack Ma-backed firm has developed techniques that slash computing costs by 20%, achieving results comparable to those of Nvidia’s (NVDA) chips. This breakthrough, detailed in a research paper released earlier this month, underscores Ant’s strategic pivot toward cost-efficient AI development, utilizing lower-cost hardware to train models via the Mixture of Experts (MoE) approach—a method gaining traction for its ability to maximize efficiency with reduced computational power.
The company’s innovative approach integrates chips from domestic giants like Alibaba and Huawei alongside U.S.-sourced alternatives from Advanced Micro Devices (AMD), reducing its historical reliance on Nvidia, a leader in AI hardware. Bloomberg sources note that while Ant continues to employ Nvidia chips, the shift toward Chinese semiconductors and AMD options reflects a deliberate effort to diversify supply chains and mitigate risks tied to dependence on a single provider. This move aligns with broader industry trends, particularly as the U.S. imposes restrictions on China’s access to cutting-edge semiconductors, limiting availability of Nvidia’s most advanced offerings while still permitting sales of lower-end chips. Despite these constraints, Ant’s ability to maintain performance parity with Nvidia’s technology – evidenced by its 20% cost reduction – highlights China’s growing prowess in adapting to geopolitical challenges.
On Monday, Ant unveiled “major upgrades” to its AI-driven healthcare solutions, now deployed across seven major hospitals and institutions in cities like Beijing, Shanghai, Hangzhou, and Ningbo. Built on a foundation of DeepSeek’s R1 and V3 models, Alibaba’s Qwen, and Ant’s proprietary BaiLing, this healthcare-specific AI can address medical queries and enhance patient services, according to the company’s statement. The announcement coincided with Bloomberg’s report, amplifying Ant’s narrative of technological resilience. Meanwhile, Nvidia’s stock, trading at $120.13, shrugged off the news with a 2.06% uptick, suggesting market confidence in its broader dominance despite Ant’s advancements.
Ant’s dual-chip strategy not only trims training time and costs but also positions it as a formidable player in a global AI race increasingly shaped by efficiency and self-reliance. The MoE technique, which segments tasks across multiple networks, allows Ant to extract high performance from less resource-intensive hardware—a critical advantage as China navigates U.S. export controls. It goes without saying that Ant’s actions speak volumes. By blending domestic innovation with select foreign technology, the company is carving a path that could redefine AI economics. With Alipay’s vast user base providing a real-world testing ground, Ant’s healthcare AI rollout exemplifies how these cost-saving techniques can translate into practical, scalable solutions, potentially setting a precedent for other Chinese tech firms facing similar semiconductor constraints.
WallStreetPit does not provide investment advice. All rights reserved.
Leave a Reply