Alibaba Entertainment Mid‑Platform Strategy: Insights from Senior Director Wei Zhou

In the past six months, the concept of a "mid‑platform" (中台) has attracted great attention in the Chinese tech industry. Alibaba first introduced the "big middle platform, small front‑end" idea in December 2015, aiming to build an agile front‑end combined with a powerful middle platform. Since then, companies such as Tencent, Baidu and JD.com have also begun reorganizing to prepare for their own mid‑platform initiatives.

The interview with Wei Zhou, senior director of Alibaba’s Entertainment Group and head of the foundational platform, explores why and how Alibaba’s entertainment business is adopting a mid‑platform strategy, and what value it can bring.

Evolution of Alibaba’s Mid‑Platform

Wei Zhou outlines four stages of Alibaba’s business‑mid‑platform construction:

Single‑system stage – a single business system runs on a few machines, gradually moving to a Java stack.

Distributed‑system stage – large teams are split into smaller, high‑cohesion, low‑coupling systems (user center, product center, transaction center).

Business‑center platform stage (2011) – business logic is abstracted and modularized, solving 80% of common problems through abstraction and 20% of specialized needs via open architecture.

Business‑mid‑platform stage – as complexity grows, a dedicated mid‑platform is created to reduce information‑acquisition cost, inter‑connectivity cost, and duplicated development.

Because the entertainment business differs significantly from Alibaba’s e‑commerce core, the entertainment mid‑platform cannot simply copy the group’s platform; it must adapt the shared capabilities (data, middleware) to its own needs.

Entertainment Mid‑Platform Strategy

The entertainment group consists of many acquired businesses (e.g., Maimai, UC, Youku) with heterogeneous back‑ends. To achieve a unified platform, the group first focuses on “core‑business anchors.” Youku and UC are identified as the primary anchors, with Youku’s video, advertising, and membership services driving the initial mid‑platform development.

Technical migration examples include:

Original Youku CDN and web services were built on a custom Linux kernel and later integrated into Alibaba Cloud.

In 2010, a mobile CMS was rapidly built with Ruby, but later rewritten using Facebook’s Tornado framework (Python + C) and eventually re‑implemented in Java to align with the group’s Java‑centric stack.

Similar re‑engineering efforts were undertaken for other subsidiaries (e.g., ticketing platforms, literature services, film‑theater management).

Key Challenges and Recommendations

Wei Zhou emphasizes that a mid‑platform must quickly deliver value to core businesses; otherwise, the strategy stalls. He also notes that mid‑platform adoption is more suitable for larger, information‑intensive enterprises (finance, automotive, telecom) rather than very small startups.

Practical advice includes assessing whether an organization has:

Comprehensive data and resource systems in place.

Many internal requests that can be consolidated.

The need for common tools across teams.

The ability to expose core functions as services.

These criteria correspond to the four maturity stages: systematization, centralization, platformization, and mid‑platformization.

Beyond data integration, a mid‑platform enhances development agility and enables rapid response to business changes, representing an innovative organizational architecture rather than a simple data hub.

Future Outlook

Wei Zhou envisions a fully unified entertainment mid‑platform that supports all subsidiaries, while still allowing vertical teams to retain unique capabilities (e.g., paid‑reading models for literature versus ad‑supported video for Youku). Emerging technologies such as 5G, AI, VR/AR, and computer vision are also being explored to improve user experience and content delivery.