11/20 2026
443
By Yang Jianyong
Powered by artificial intelligence, the global cloud services market is demonstrating robust growth momentum. In the third quarter of this year, worldwide enterprise spending on cloud infrastructure services once again shattered previous records, reaching $168 billion—a 55% year-over-year increase.

It is worth noting that before ChatGPT's launch in late 2022, global cloud development had entered a bottleneck period, with growth steadily declining. Since AI ushered in the generative AI era, however, the global cloud market has expanded 185%, with the year-over-year growth rate tripling.
Amazon, Microsoft, and Google—the world's three major giants—dominate the global market. Among them, Amazon's AWS cloud service continues to maintain a strong leading position, with revenue exceeding $160 billion (approximately RMB 1.06 trillion) over the past 12 months and a backlog of $578 billion in undelivered orders. Amazon's CEO optimistically stated that AWS revenue is expected to reach $1 trillion. As the world's second-largest public cloud vendor, Microsoft's Azure annual revenue exceeded $100 billion for the first time.
Driven by AI demand, enterprise appetite for AI infrastructure and AI solutions remains strong. In particular, the continuous improvement of large model capabilities and AI agent applications is ushering in a new stage, further propelling the global cloud services market's strong growth momentum. Against this backdrop, Alibaba Cloud has also set a $100 billion revenue target.
In the second quarter of fiscal year 2027 (the natural quarter being the third quarter of 2026), Alibaba Cloud's single-quarter revenue broke through the RMB 60 billion mark, reaching RMB 60.973 billion, up 53% year-over-year. Revenue from external customers grew 64% year-over-year.
Leveraging the full-stack AI advantages formed by chip-cloud-model integration, Alibaba Cloud demonstrates strong competitiveness in the domestic cloud market, ranking first in the AI cloud market. AI-related product revenue this quarter reached RMB 17.918 billion, marking the thirteenth consecutive quarter of triple-digit year-over-year growth.
In the high-end chip sector, T-Head's three-generation GPU product roadmap is accelerating. As of May 2026, the Zhenwu series chips have cumulatively shipped 560,000 units. This year, the Zhenwu M890 will be fully launched, and the next-generation training-inference integrated Zhenwu V900 AI chip is planned for launch in the first quarter of 2027.
Most critically, T-Head has achieved comprehensive coverage of the core chip layout for hyperscale AI clusters—including Zhenwu GPU chips, Yitian CPU chips, Panmai smart NICs, ICN Switch interconnect chips, and Zhenyue enterprise-grade storage controllers—achieving full-stack in-house development of computing power, network power, and storage power.
Self-developed AI chips have become the supporting foundation of Alibaba's full-stack AI technology, providing workloads for itself and Alibaba Cloud customers in model training, model inference, and Agentic tasks—the core foundation driving AI revenue growth.
At the AI model layer, the Qwen large model demonstrates leading positions in reasoning, programming, and other areas, and has also launched video generation models and world models to enrich the multimodal model matrix. Among them, the latest release Qwen4-Max, based on innovative model architecture and training optimization, achieves higher performance with fewer Token consumption, continuously evolving in key capability areas such as agents and programming.
To date, Alibaba has open-sourced more than 470 large models, encompassing text generation models, visual understanding/generation models, speech understanding/generation models, text-to-image and video models, covering all size parameters from 0.5B to 2.4T across various tiers. Globally, the Qwen series models have cumulative global downloads exceeding 4 billion, with over 300,000 derivative models, firmly ranking as the world's No. 1 open-source model family.
Currently, AI computing power remains in short supply globally. However, due to new global computing power export control regulations, Alibaba Cloud's chip supply will primarily rely on T-Head, while also purchasing domestic chips to meet AI infrastructure needs. But changes in the external supply chain have impacted Alibaba Cloud's data center construction, and the previously set target of exceeding 20GW in data center scale by 2032 has been lowered accordingly.
Affected by this, Alibaba Cloud's growth will slow, with growth in the next fiscal year slowing from the current 53% to a range of 30% to 45%. It also noted that with the continuous expansion of domestic computing power product scale, high-speed growth of over 50% is expected to resume only in 2028. At the same time, the company stated that the target of $100 billion in external revenue and over 20% profit margin for Alibaba Cloud in the 2030 natural year remains unchanged.
Finally, demand for AI infrastructure remains robust, coupled with the increasing maturity of agent development systems and Tokens becoming cheaper. With the continuous iteration of AI Agents, increasingly powerful capabilities, and the background of gradually declining Token costs, Token usage will be further driven. Under this trend, it will help boost revenue growth for cloud computing companies and unleash long-term growth potential.
The views expressed represent personal opinions only. Focused on in-depth observation of technology fields such as large models, artificial intelligence, robots, IoT, cloud computing, and smart home appliances.