Wukong: The AI Platform Built for the Office

Here's what Alibaba is betting on: that the future of enterprise work isn't one AI doing one thing — it's multiple AI agents working together, coordinated through a single interface. That's the core idea behind Wukong, the company's new AI-native enterprise platform, unveiled in March at the AI DingTalk 2.0 launch event.

Named after the Monkey King from the classic Chinese novel Journey to the West, Wukong handles the kind of work that fills most people's days — document editing, approval workflows, meeting transcriptions, research. And it does it by orchestrating multiple AI agents at once. Right now it's invitation-only beta, but it already runs as a standalone app or through DingTalk, Alibaba's cloud-based workplace tool that already has more than 20 million corporate users. That's not a small sandbox to test in.

Enterprise Security and a Bigger Ecosystem Play

Alibaba has been clear that Wukong isn't just a clever demo. The company says it comes with enterprise-grade security infrastructure, and there are plans to connect it with tools people are already using — Slack, Microsoft Teams, Tencent's WeChat. And longer-term? It's getting woven into Alibaba's broader world: Taobao, Alipay, the works.

Wukong now sits under a newly formed business group called Alibaba Token Hub, led directly by CEO Wu Yongming. That's not a small signal. When a CEO-level unit takes ownership of your agentic AI strategy, it means this isn't a side project.

A New Flagship Model Drops on April 20

On top of the platform push, Alibaba released Qwen3.6-Max-Preview — described as its most powerful language model yet. Compared to the recently released Qwen3.6-Plus, it brings improvements in agentic coding, world knowledge, and instruction-following. Think of it as the top of a lineup that's growing fast.

The Qwen3.6 Family Is Expanding Quickly

The Qwen3.6 series now includes Qwen3.6-Flash, Qwen3.6-Plus, and the open-source Qwen3.6-35B-A3B, which landed on Hugging Face in mid-April with a specific focus on agentic coding and thinking preservation. There's a clear pattern here: Alibaba isn't releasing one model and waiting — it's building out an entire ecosystem of models at different capability and cost levels, targeting developers and enterprises alike.

Wall Street Is Starting to Pay Attention

Here's where it gets interesting from an investment angle. Bernstein estimates Alibaba's AI spending in the March quarter nearly doubled from the prior quarter — landing at roughly 20 billion yuan, or about $2.93 billion. That's a massive ramp. And management has set a five-year revenue target of over $100 billion combined from cloud and AI, backed by a commitment of at least 380 billion yuan — around $53 billion — in AI and cloud infrastructure over three years.

Analysts Are Upgrading and Raising Targets

Bank of America reiterated a Buy rating with a $180 price target. Analyst Joyce Ju noted that the scale of reinvestment had been "meaningfully underestimated" — which, honestly, is a polite way of saying the market wasn't really pricing this in.

Goldman Sachs went further, upgrading Alibaba to Conviction Buy and projecting an earnings recovery by fiscal 2027 and 2028, driven specifically by AI and cloud momentum. Barclays trimmed its price target slightly — from $190 to $186 — while keeping an Overweight rating, citing near-term pressure from heavy infrastructure spending. The consensus target still sits in the $185–$190 range, implying around 35% upside from recent levels.

Alibaba's Hong Kong-listed shares rose more than 14% in April through mid-month — on pace for the strongest monthly gain since January. That kind of move doesn't happen without conviction building behind the scenes.