Alibaba positions Qianniu to drive e-commerce growth

Starting April 1, 2026, Alibaba has reorganized its e-commerce AI strategy to place Qianniu at the center of growth and to channel commercialization through a newly created Alibaba Token Hub (ATH). The move consolidates AI responsibilities and ties product roadmaps to a token-based operating model led by group leadership.

Qianniu is being repositioned from a product suite into an agentic AI platform that can drive B2B merchant tools and marketplace orchestration. ATH will centralize multimodal and algorithmic capabilities (previously housed in the future innovation unit) and act as the primary commercialization engine. The group is explicitly shifting success metrics: from B2C user-side penetration and search/UX gains toward merchant-centric KPIs such as retention of AI-driven merchant tools and gross merchandise value (GMV) attributable to AI.

Organizationally, the smart search and recommendation products division was split into two units: platform users & products, and algorithms. Kaifu Zhang, who formerly led AI efforts in the domestic e-commerce business group, no longer holds that role. OpenClaw — an internal initiative — has accelerated interest in agentic AI, and Eddie Wu has created ATH and directed all related businesses to commercialize around it. Earlier B2C efforts under Zhang produced features like universal search, a product selection assistant, shopping guides, and scan-to-search to improve user experience; the new phase prioritizes B2B advertising performance and merchant growth.

This is a live example of a major e-commerce platform moving from incremental model improvements to platform-level agentic capabilities and a token-aligned operating model. For ML engineers and product teams, expect increased emphasis on agent interfaces, merchant-facing APIs, retention optimization, model-in-the-loop ad systems, and measurement frameworks that tie model outputs directly to GMV. For researchers, ATH’s consolidation of multimodal work could accelerate productionization of agentic and multimodal models at scale.

Implementation details: how Qianniu’s agentic APIs are exposed to merchants, the token mechanics ATH uses to incentivize behavior, the metrics and attribution methods for AI-driven GMV, and any public SDKs or partner programs. Also watch team leadership announcements and whether ATH publishes technical roadmaps or opens interfaces for third-party developers.