Alibaba’s artificial intelligence growth is set to accelerate in the second half of 2026, the Chinese technology group said, as it announced a series of organisational, product and infrastructure milestones from the first half of the year that it says lay the groundwork for the next phase of enterprise AI adoption.
Rebuilding Around AI: The Alibaba Token Hub
The most significant move came in March, when Alibaba established the Alibaba Token Hub (ATH) Business Group directly under Chief Executive Officer Eddie Wu, who took the helm in September 2023 and is a co‑founder of the company. ATH consolidates five previously separate units—Tongyi Laboratory, the MaaS (Model‑as‑a‑Service) Business Line, the Qwen Business Unit, the Wukong Business Unit, and the AI Innovation Business Unit—under a single organising mission: to create, deliver, and apply tokens. The restructuring, which followed Alibaba’s broader reorganisation into six independent business groups in March 2023, is designed to bring every core AI team and product together in one umbrella, giving the company a unified front as enterprises shift from early‑stage experimentation toward wider AI deployment.
Alibaba now emphasises full‑stack AI capabilities that span proprietary chips (through its T‑Head subsidiary), cloud infrastructure, foundation models and applications. That integration, the company argues, gives it a structural advantage in the accelerating market. The bet is already showing in revenue: AI‑related product revenue for the Cloud Intelligence Group has delivered triple‑digit growth for multiple consecutive quarters, and MaaS token revenue grew fifteen‑fold year‑on‑year in the first five months of 2026. The heavy investment, however, has put pressure on short‑term profitability—adjusted EBITA declined significantly in the September quarter of 2025—though Alibaba maintains a strong liquidity position.
Advanced Models and an Expanding Ecosystem
On the model front, Alibaba released Qwen3.7‑Max in May, a next‑generation large language model that the company says delivers advanced agentic coding, complex reasoning and long‑horizon task execution. According to analysis firm Artificial Analysis, it outperforms leading Chinese models and matches top global systems. The Qwen family remains the world’s most widely used open‑source model family, surpassing one billion cumulative downloads on Hugging Face as of January 2026; an earlier version, Qwen3.5, was launched in February and showed strong performance in reasoning, coding, agentic tasks and multimodal understanding.
Alibaba also pushed into generative video with HappyHorse 1.1, released in June, which brings improved motion realism, consistency and visual quality. Its predecessor, HappyHorse 1.0, debuted in April and has since been widely adopted across short‑form content, advertising, brand marketing and gaming cinematics. In the same month the company introduced HappyOyster 1.0, an interactive world model that offers enhanced environmental interaction, expanded controls and rewindable storylines, enabling faster production of interactive films, games and visual experiences.
The ecosystem around these models is broadening. In January the Qwen consumer AI assistant was upgraded to integrate services from Taobao, Alipay, Fliggy and Amap into a single conversational interface, allowing users to handle shopping, payments, travel and navigation seamlessly. At MWC Barcelona, Alibaba unveiled Qwen Glasses, which provide real‑time translation, HD capture, transcription, visual recognition and payments; in China users can already order food and hail rides through voice commands. In March the Wukong Platform was introduced as an enterprise‑grade agentic platform for complex, multi‑step workflows, serving as Alibaba’s primary solution for autonomous enterprise operations. Separately, the international agent Accio Work—designed to automate compliance, marketing and sourcing for small and medium‑sized enterprises—is being brought to UK manufacturers.
Scaling Global Infrastructure and Industry Impact
Underpinning these efforts is a US$53 billion AI infrastructure commitment announced in February 2025. Alibaba spent roughly US$16 billion on AI‑related procurement, including data centres, in the 12 months to July 2026. Its global cloud footprint now reaches 105 availability zones across 32 regions, after new data centres came online in Japan, Malaysia, France and Mexico. The company already had a significant European presence: it launched its first UK data centres in London in October 2018, with two availability zones, and has since opened a Paris hub—its third European site alongside Germany and the UK. Reports indicate Alibaba is also seeking to contract 50 megawatts of power for a planned data centre in Ireland. In February 2025 the group elevated overseas expansion to a strategic priority, creating a dedicated overseas business unit. The new data centres deliver enterprise‑grade cloud computing services designed to comply with local cybersecurity, resilience and data governance rules, with a stated priority on data privacy and sovereignty.
Alibaba has also pursued partnerships to deepen its market reach. It has held previous talks with BT Group Plc about a cloud services partnership, and in early 2026 entered a strategic alliance with Systango, a global AI and digital engineering firm, to accelerate AI adoption in the UK and other markets.
The technology is now being applied in critical sectors. In early 2026, Alibaba DAMO Academy launched two AI‑powered screening tools: MAOSS for early detection of fatty liver disease, and COCA for colorectal cancer screening. Both achieved high diagnostic accuracy; COCA reportedly demonstrated higher sensitivity in detecting colorectal cancer from CT scans than radiologists. In May, Alibaba Health launched “Hydrogen Ion”, a medical AI assistant for healthcare providers, developed in partnership with the UK’s BMJ Group.
In agriculture, Alibaba Cloud has partnered with Muyuan Group, a global leader in livestock farming, to build an intelligent swine‑farming AI model using the Qwen large language model. The project aims to accelerate AI applications in feed nutrition, breeding stock improvement and livestock management. The initiative has drawn scrutiny: one thinktank has warned that large technology firms, including Alibaba, Google, Microsoft and Amazon, are “playing with the food system” by using AI to influence crop choices, potentially undermining farmers and concentrating corporate control.
Alibaba has also faced data privacy concerns more directly. In April 2026, UK Biobank participant data—though de‑identified—was listed for sale on Alibaba e‑commerce platforms in China. The company swiftly removed the listings, but the incident highlighted ongoing debates about AI data ethics, privacy safeguards and the need for adaptive regulation. The UK government’s approach to AI regulation, set out in a March 2023 white paper, emphasises balancing innovation with risk mitigation through principles of safety, transparency, fairness, accountability and contestability, and the UK has established the Artificial Intelligence Safety Institute (AISI). Alibaba itself has adopted six core AI ethics principles—human‑centric, inclusive, safe, reliable, private, trustworthy, accountable and open to collaboration—and invested in privacy‑preserving computation, explainable AI and data security.
Alibaba Cloud now ranks as the world’s fourth‑largest cloud provider, behind Amazon, Microsoft and Google, with a strong position in the Asia‑Pacific region. The company’s workforce has shrunk by approximately 34 per cent year‑on‑year as of December 2025, partly due to the exit from offline retail businesses and a strategic shift toward less labour‑intensive, AI‑focused areas.
