Alibaba Group Holding shares surged to a near four-year high in Hong Kong trading as Goldman Sachs raised its price target on the Chinese e-commerce giant, citing a revival in China's artificial intelligence infrastructure narrative driven by explosive enterprise AI adoption.

The investment bank lifted its 12-month target price for Alibaba to 179 from 163, maintaining a "buy" rating as enterprise-level large language model token consumption in China soared 363% in the first half of 2025. Goldman highlighted Alibaba's leading position in China's public cloud market and diversified chip supply strategy as key advantages in the rapidly expanding AI infrastructure sector.

Hong Kong-listed Alibaba shares opened 2.74% higher on Tuesday, with the company's market capitalization returning above HK3trillion($384 billion). The stock has gained more than 94% year-to-date, outpacing broader market gains as investors bet on China's AI transformation.

Goldman's upgraded outlook reflects growing confidence in Chinese cloud providers' ability to capitalize on surging enterprise AI demand, with the bank projecting 39% year-over-year growth in capital expenditure for Chinese cloud service providers in the third quarter of 2025.

Enterprise AI Adoption Accelerates Token Consumption

Chinese enterprises are embracing generative AI at unprecedented rates, with daily token consumption by enterprise-level large language models reaching 10.2 trillion in the first half of 2025, according to Frost & Sullivan data cited by Goldman Sachs. This represents a 363% surge from the second half of 2024.

Alibaba has emerged as the top choice among Fortune 500 Chinese companies deploying generative AI in 2025, according to Omdia research referenced in the Goldman report. The company's penetration in this segment, combined with its revenue model based on API calls and token usage, positions it favorably as enterprises scale their AI implementations.

Among model providers, Alibaba, ByteDance, and DeepSeek rank as the top three options for Chinese enterprises selecting general-purpose large language models, the data showed.

Multi-Chip Strategy Reduces Supply Dependencies

Goldman analysts, led by Ronald Keung, emphasized that Chinese cloud providers have made progress in developing proprietary inference chips while adopting "multi-chip strategies." This evolution means China's AI cloud industry growth "is no longer solely dependent on overseas chip supply," according to the report.

The diversified chip supply approach, combined with robust capital expenditure prospects, creates compound growth potential for the sector. Goldman projects Chinese cloud service providers' capital spending will increase 39% year-over-year in the third quarter, providing hardware foundation for sustained AI cloud revenue growth.

Alibaba holds a 47% share of China's public cloud market according to IDC data, with Goldman noting the company's "leading model capabilities" and potential for international expansion. The investment bank specifically raised its valuation for Alibaba Cloud to 43 per American Depositary Share from 36, while upgrading growth expectations for the cloud division to 30%-32% for fiscal 2026's second through fourth quarters.

Infrastructure Investment Narrative Revives

Goldman draws parallels between China's current market phase and the U.S. market two years ago when ChatGPT launched, noting that investors then rewarded companies that increased capital expenditure under the premise of "build it and they will come." The bank believes a similar dynamic is emerging in China as enterprises rapidly adopt AI technologies.

Recent model developments underscore the competitive landscape, with Alibaba releasing its Qwen3-Next architecture on September 12, claiming 10-fold performance improvements over previous generations at one-tenth the building cost. The 80-billion parameter Qwen3-Next-80B-A3B model runs 10 times faster than the 32-billion parameter model released in April.

Despite these technological advances, monetization remains challenging. As of August 2025, global AI applications generated approximately 30 billion in annual recurring revenue, while Chinese AI applications contributed only 1.5 billion, representing just 5% of the global share. Goldman noted that while Chinese models are rapidly closing the gap with global peers in multimodal areas like text-to-video and text-to-image generation, Chinese AI applications still face a long road to commercial success.