Jan 15, 2025 · 24-Hour AI Briefing: Qwen Hits 100M MAU in Two Months by Plugging Into Daily Life, and Microsoft Spends ~$500M/Year on Anthropic to Go Multi-Engine
Two updates from different poles of the AI market, but the same underlying shift: the battle is moving from “best model” to “default entry point + reliable execution in real workflows.” Alibaba is turning AI into an action-oriented life assistant by wiring Qwen into its ecosystem. Microsoft is strengthening platform resilience and leverage by betting on a multi-model “multi-engine” architecture.
1. Qwen crosses 100M MAU in two months: deep integration with Taobao, Alipay, Fliggy, Amap, and more; opens AI ordering/shopping/booking to all users
Commentary:
Reaching 100M MAU in two months is a distribution story as much as a product story. It highlights Alibaba’s ability to route traffic and convert ecosystem touchpoints into adoption. More importantly, the integration with Taobao/Alipay/instant commerce/Fliggy/Amap moves Qwen from a chat interface toward an action-capable life assistant—one that can actually place orders, buy things, and book travel.
Opening the feature set broadly (rather than a narrow gray test or paid tier) signals strategic intent: accelerate iteration with real demand, capture user habit early, and lock mindshare.
In China’s crowded field (Doubao, DeepSeek, Baidu’s Ernie, etc.), Alibaba remains in the top tier. The ceiling will be set less by flashy features and more by task completion rate, robust fallback flows, and user trust in letting an agent handle commerce and travel end-to-end.
2. Microsoft becomes a top Anthropic customer, reportedly spending ~$500M/year to power Microsoft products
Commentary:
In a platform-level AI race, Microsoft can’t tie the AI layer of Office, Windows, GitHub, Azure, and Security to a single vendor or single roadmap. Working with multiple model providers is fundamentally about building a “multi-engine” architecture—redundancy plus negotiating leverage.
As OpenAI’s independence grows and strategic tensions rise, Microsoft needs optionality. The likely outcome isn’t a wholesale engine swap, but scenario-based routing: some Copilot capabilities on model A, others on model B, and Azure positioning multi-model choice as a standard enterprise commodity. That expands monetization from “selling models” to “selling the platform that orchestrates them.”
Until AGI arrives, the most defensible winners may be platforms that integrate, schedule, and monetize diverse AI capabilities—not merely the labs with the strongest single model.
Closing:
Alibaba is pushing AI as a consumer “do-things” entry point across daily life. Microsoft is pushing AI as a modular, swappable platform layer. Which approach do you think compounds faster: ecosystem-driven super-app entry points, or multi-engine platform orchestration?
Further reading (top AI events in the last 72 hours):