Alibaba Unifies E-Commerce Groups and Integrates Cloud-Chip Assets in Sweeping Restructuring

The market often views Alibaba's latest earnings—a 75% plunge in net income and a 30% drop in adjusted earnings to 27.33 billion yuan ($4.1 billion)—as a sign of a struggling retail giant losing ground to upstarts. The reality, buried in a sweeping corporate reorganization announced Thursday, is that Alibaba is intentionally cannibalizing its short-term profits to fund the largest private AI infrastructure buildout in Chinese history.[1][2][3]

Alongside its fiscal first-quarter 2027 results, the company unveiled a radical consolidation of its sprawling empire into three core pillars. The most consequential move merges its Cloud Intelligence Group with its in-house semiconductor design firm, T-Head, creating a unified "AI Cloud and Compute Services" division.[1][4]

By fusing silicon design with cloud operations, Alibaba is building a vertically integrated AI stack—from the chip to the server to the model. This structural shift treats the semiconductor, the network, and the cloud as a single economic unit rather than separate businesses.[3][4]

This is a direct response to U.S. export controls that restrict access to advanced Nvidia hardware. T-Head's proprietary Zhenwu accelerators, including the M890, are now being deployed at scale in massive supernodes to train Alibaba's Qwen large language models. The company reported that over 650 external customers across 20 industries are already utilizing Zhenwu-family accelerators.[3][5]

This vertical integration is extraordinarily expensive. Capital expenditures surged 75% year-over-year to 67.68 billion yuan ($9.9 billion) in the June quarter. The cost of deploying new servers has roughly doubled year-over-year due to component scarcity and the massive compute requirements of AI agents.[1][2][3]

CEO Eddie Wu confirmed the company has already spent half of its planned 380 billion yuan ($56.4 billion) AI investment budget for the 2026–2029 period. The strategic bet is that owning the underlying compute will eventually yield substantially higher gross margins, insulating Alibaba from the volatile pricing of commercially procured chips.[2]

While the cloud segment builds the infrastructure, the retail segment is being rewired to fund it. The restructuring dismantles the long-standing wall between Alibaba's domestic and international retail operations. The company merged its China E-Commerce Group (Taobao and Tmall), its International Digital Commerce Group (AliExpress, Lazada), and its grocery chain Freshippo into a single "Alibaba E-Commerce Group."[1][4]

This unified front is designed to counter fierce competition from Pinduoduo and ByteDance's Douyin, which have steadily eroded Alibaba's market share. By pooling logistics, merchant data, and AI-driven marketing tools across borders, the company aims to extract cross-platform synergies that were previously blocked by internal silos.[4]

To bridge the gap between infrastructure and end-users, Alibaba consolidated its scattered AI consumer and enterprise applications—including the Qwen model labs and QwenWork—into a dedicated "AI Labs and Applications" segment. This unit integrates the full value chain from AI model innovation through to consumer applications and enterprise productivity solutions.[1][4]

Building the entire AI stack remains a massive financial drain. While the newly formed AI Cloud and Compute Services segment posted a positive adjusted EBITA of 5.63 billion yuan, the AI Labs and Applications unit generated a steep 13.86 billion yuan adjusted EBITA loss for the quarter. That single-segment loss is more than double the profit produced by the cloud and compute division.[3]

The primary uncertainty is whether Alibaba's in-house silicon can truly match the performance of next-generation foreign hardware over the long term. While the Zhenwu chips provide a vital lifeline amid sanctions, training frontier models requires exponential increases in compute power. If the proprietary hardware falls behind, Alibaba risks burning billions on an inferior stack.[3][5]

The reorganization marks a definitive end to Alibaba's era of decentralized expansion. By tying its e-commerce cash engine directly to an integrated cloud-and-chip infrastructure, the company is betting that full-stack AI capabilities will be the ultimate moat in China's tech sector, even if it means enduring quarters of compressed margins.[1][3]