Xiaoice, Microsoft's Chinese chatbot, has a sense of humor, good listening skills, and 20M registered users
For Sympathetic Ear, More Chinese Turn to Smartphone Program — She is known as Xiaoice, and millions of young Chinese pick up their smartphones every day to exchange messages with her …
Context & Ripple Effects
The NYT profile captures Xiaoice at the start of its arc: in mid-2015 Microsoft's China-built chatbot has 20M registered users who use it as a sympathetic ear rather than a utility, making emotional companionship — not task automation — the product. What followed validates the bet: by 2018 the bot had graduated from text to placing human-sounding phone calls to people across Chinese platforms, prefiguring Google's Duplex.
Microsoft then cut it loose: in 2020 it spun the five-year-old bot into an independent Shanghai-based company while retaining a stake, and a year later the standalone Xiaoice raised $77M at a $1B valuation claiming over 660M users. The story matters because it marks the moment conversational AI proved mass-market pull in China years before Western assistants found comparable engagement.
First-order effects
- Microsoft holds both proof points competitors lacked in 2015 — real daily engagement at 20M-user scale in China and a monetizable asset — which it converts to equity value via the 2020 spin-out while keeping a stake.
Second-order effects
- Xiaoice's independence turns it into a competitor to the platforms that hosted it, pushing China's internet giants toward voice-first distribution channels like the smart speakers they embraced in 2019 to reach newly connected users.
Third-order effects
- By the time SenseTime demos SenseChat in Chinese and English on its own LLM in 2023, the market Xiaoice seeded — consumer-facing conversational AI in China — has become contested ground between spun-off pioneers and foundation-model entrants.
The trend: Consumer AI companionship, proven at scale by Xiaoice in China, is evolving from a corporate side project into standalone venture-backed companies competing against LLM-native entrants.