The Memo That Ended a Winner
On 17 August 2026, Alibaba told its own staff first. Zhou Bingshu, head of the Lingxi Games studio, sent an internal memo announcing that Alibaba was divesting its entire stake in the unit to Trustar Capital, an Asian private equity firm, for a deal reported at a minimum of 1.5 billion US dollars - Reuters puts the figure Alibaba could ultimately receive above 2 billion. Lingxi's one real hit is Three Kingdoms: Strategy Edition, a mobile strategy game built with Japan's Koei Tecmo that has quietly generated steady revenue in a gaming market most large platforms treat as a distraction from their main business. Neither Alibaba nor Trustar Capital commented on terms beyond the memo.
Zhou's stated reason was that the parent company needed to 'better focus on its strategic priorities' - a phrase that names no flaw in Lingxi's performance, because there is not one. This was not a struggling unit being cut loose. It was a working, monetizing business sold to fund something else.
Why It Matters: The Bet Is Still Unproven
The something else is Alibaba's AI pivot under chief executive Eddie Wu, who has set an internal target of 100 billion US dollars in AI revenue within five years. Alibaba has also recently promoted what it describes as the largest AI model it has ever produced, with performance the company says rivals Anthropic's latest systems. Both claims are Alibaba's own framing - neither the revenue target nor the model comparison has been independently verified, and a five-year revenue target is, by definition, not yet money in the door.
Selling Lingxi does not fund a proven business at a known return - it funds a bet on hitting a number that is still four-plus years away, made by a company that has not yet shown the AI revenue to match the ambition. That gap between the certainty being given up and the certainty being bought is the actual decision, and it is the part the internal memo does not mention.
Yes, But: Sometimes the Cash Cow Has to Move
None of this makes divesting automatically wrong. A capital-intensive AI buildout genuinely can need cash faster than a gaming studio's earnings can generate it, and a company with a hard five-year target may reasonably prefer certain capital now over slower reinvested profit later. Concentrating a scattered portfolio around one strategic bet is a legitimate allocation choice, not just a rationalization - plenty of well-run companies have sold a good business to fund a better one and been proven right in hindsight.
The distinction that matters is whether the trade was actually evaluated as a trade - proceeds now versus the studio's ongoing earnings power, discounted for the AI bet's real odds - or whether 'strategic priority' was used as a label that made the comparison unnecessary. Zhou's memo offers no sign the first analysis happened; it reads like the second.
The Bottom Line: Price the Trade, Not the Label
Before any business unit gets reclassified as 'non-core' to feed a strategic priority, force the same test onto the page: what is this unit's proven, ongoing earning power, and what has the priority it is funding actually delivered so far, not what it is targeted to deliver in five years. A working, profitable unit and an unproven bet are not directly comparable until you have discounted the bet by its real probability of hitting its own target - and if nobody in the room can produce that number, the label 'strategic priority' is doing the work an actual valuation should be doing.
Alibaba may turn out to be right that AI outgrows gaming by enough to justify the trade. What should worry any owner watching is not that call - it is how easily 'strategic priority' let the trade happen without anyone having to prove it first.
Read next: Qwen's 3 Billion Downloads Hide a Compliance Gap | China Lets Apple Run Its Own AI Model, A First
