Latest Cover Story | Two Giants Compete for Position in AI (AI Translation)

最新封面报道｜双雄卡位AI

1. Alibaba and Tencent, two of China's largest internet companies, are entering a new phase of significant investment, aiming to seize opportunities in AI large model development and AI applications, similar to global technology peers. In early 2024, the emergence of DeepSeek, an AI company founded by Liang Wenfeng, rapidly gained worldwide attention. Both Alibaba's Jack Ma and Tencent's Pony Ma attended discussion panels with Liang. Notably, Tencent integrated DeepSeek’s R1 foundational model into its AI chatbot Yuanbao and paused some training of its own Hunyuan model to prioritize DeepSeek integration. Subsequently, Alibaba CEO Wu Yongming announced a record investment of over RMB 380 billion (~$53 billion) over the next three years towards cloud and AI infrastructure—more than its total investment in the previous decade, setting a record for private investment in these sectors in China. [para. 1][para. 2][para. 3][para. 4][para. 5]

2. DeepSeek’s success disrupted conventional thinking that only massive resource holders could participate in foundational model development; now, firms with even modest resources (e.g., 2,000 GPUs) could engage. This shift led to rapid updates and integration of external models into established AI products, with Tencent’s Yuanbao delivering 36 updates in four months and briefly surpassing DeepSeek in downloads, maintaining a top-three position alongside ByteDance’s Doubao and DeepSeek itself. Meanwhile, Alibaba’s Qwen model series offered cost-efficient, parameter-varied open-source models, enjoying rapid adoption—by February 2024, over 100,000 open-source derivatives existed, with major releases every six months. These developments illustrated two diverging strategies: Alibaba prioritized ecosystem-building through open-source models and cloud MaaS (Model as a Service), whereas Tencent accelerated integration of powerful AI functionality into billion-user applications like WeChat. [para. 6][para. 7][para. 8][para. 9][para. 10][para. 11][para. 12][para. 13][para. 14][para. 15][para. 16][para. 17][para. 18]

3. Despite different focuses, both companies are committed to full-scale AI transformation. Alibaba’s strategic pillars are infrastructure construction, foundational platform and application upgrades, and AI-driven transformation across existing businesses, while Tencent invests heavily in models, agents, and knowledge bases. Both have accelerated cloud and hardware investments and maintained intensive participation in funding AI startups, becoming shareholders or investors in most prominent Chinese AI companies, such as Zhipu AI, MiniMax, Baichuan, and Moonshot. Tencent has also backed hardware players and chip firms. The market generally anticipates that Tencent, Alibaba, and ByteDance will likely become China's AI giants, each leveraging different strengths—Tencent and ByteDance in application scenarios, Alibaba in cloud infrastructure. [para. 19][para. 20][para. 21][para. 22][para. 23][para. 30][para. 31][para. 32][para. 33][para. 34]

4. Alibaba is China’s pioneer in adopting open-source models and the MaaS model in cloud services, seeking ecosystem-building and commercial revenue. Massive investments in model development and rapid iteration—sometimes releasing five model generations in under two years—have enabled Alibaba Cloud to attract 15 million developers and attain over 70,000 models on its ModelScope community. Cost reduction is notable: its Qwen3 model requires only 4 Nvidia H20 chips for deployment, cutting costs by 65%-75% compared to DeepSeek R1. AliCloud’s public cloud segment, focusing on AI-driven services, has rebounded to double-digit growth, with AI products fueling consecutive triple-digit revenue increases and supporting major enterprise customers, including BYD, Weibo, and state-owned enterprises. However, Alibaba faces rising competition from ByteDance’s low-cost Volcano Engine, which has rapidly escalated in large model API call market share, and from Baidu Cloud. [para. 35][para. 36][para. 37][para. 38][para. 39][para. 40][para. 41][para. 42][para. 43][para. 44][para. 45][para. 46][para. 47][para. 48][para. 49][para. 50][para. 51][para. 52][para. 53]

5. Tencent, after a January 2024 reorganization, moved its major consumer AI applications under its cloud and smart industries group, rapidly integrating DeepSeek into flagship apps. Most notably, WeChat, China’s largest social platform, piloted Yuanbao-based AI features and plans further integration with a proprietary AI agent. QQ Browser was upgraded into an AI-first product, and smart productivity tools were incubated for personalized knowledge management. Tencent’s core consumer AI apps, like Yuanbao, now see robust weekday activity, indicating rising use in workflow efficiency. Large-scale integration into WeChat and other apps is constrained mainly by compute resource availability, as surges in demand would require robust, reliable back-end infrastructure. [para. 54][para. 55][para. 56][para. 57][para. 58][para. 59][para. 60][para. 61][para. 62][para. 63][para. 64][para. 65][para. 66][para. 67][para. 68]

6. Both Alibaba and Tencent are embedding AI deeper into their core businesses, spanning advertising, e-commerce, gaming, and enterprise solutions. Tencent attributes substantial improvements in ad click-through rates to AI, and Alibaba leverages generative AI for B2B commerce, automating complex processes and customer interaction, with metrics showing AI-optimized product conversion rising by 52% and reply rates by 36%. ByteDance, meanwhile, continues its tradition of rapid application innovation, deploying high-frequency model upgrades and introducing AI-powered products across education, video, and social scenarios, and has become the largest large-model API provider based on daily call volume. [para. 69][para. 70][para. 71][para. 72][para. 73][para. 74][para. 75][para. 76][para. 77][para. 78][para. 79][para. 80][para. 81][para. 82][para. 83][para. 84][para. 85][para. 86][para. 87][para. 88]

7. High-performance AI requires vast computing power; thus, all three companies have accumulated large reserves of Nvidia and domestic chips. Alibaba, for instance, expects to possess over 53,000 Nvidia and 90,000 domestic AI chips by 2025; Tencent and ByteDance also have significant, though varying, reserves. The companies attempt to adapt to evolving US export controls by diversifying chip sources, accelerating investment in domestic chip design and heterogeneous computing clusters, but face integration challenges. Huawei is seen as a leader in domestic chip technology; Alibaba and Tencent also have chip design subsidiaries or investments but lag in independent R&D. [para. 89][para. 90][para. 91][para. 92][para. 93][para. 94][para. 95][para. 96]

8. Both Alibaba and Tencent are major AI ecosystem investors; Alibaba funds startups to extend cloud-based model and application services, while Tencent leverages its popular social platforms as integration points for third-party AI apps. Developers are attracted to Alibaba’s open-source cloud ecosystem and generous compute resources, while Tencent’s WeChat-based ecosystem offers vast user access for enterprise automation and knowledge management services. This ecosystem-centric approach is likely to shape the competitive AI landscape in China for years to come. [para. 97][para. 98][para. 99][para. 100][para. 101][para. 102][para. 103][para. 104][para. 105][para. 106][para. 107][para. 108][para. 109][para. 110][para. 111][para. 112][para. 113][para. 114][para. 115]