2026-10-10 surfaced via nichochar's timeline (X, ~11:30 AM ET)
and corroborated by Dealroom (Oct 8, citing The Information) +
Wikipedia (Dec 2025 acquisition). Chain: 2025 — launch, $75M
Benchmark raise, Singapore move, $100M ARR in 8 months, Meta
acquisition announced Dec (undisclosed; X-grade: $2B+). 2026 —
China opens a probe; 100+ staff move into Meta's office;
cofounders can't leave China (X-grade); April: government orders
the deal undone; Tencent, HSG and ZhenFund buy the shares back
from Meta; none of ~150 staff quit; Meta launches Muse; Manus
launches Cue (standalone agent app) three weeks later; a
$500M+ round announced on WeChat: led by Boyu Capital and
IDG Capital, Tencent/HSG/ZhenFund participating; The Information
(June): ARR $400–500M (up from $100M in Dec); no valuation
given (Dealroom) — the ~$4B figure is nichochar's, X-grade. The
on-record read (nichochar): the battleground is the application
layer, "China has no good application-layer players so they fight
for this one"; "Meta clearly downloaded a bunch of their thinking."
The capital note nobody else made: Boyu is CIC-affiliated — the
unwind didn't just revert a deal, it re-nationalized the cap
table.
Background
The thread: Meta's $2B+ acquisition of the Manus agent company
(Butterfly Effect), announced Dec 2025, was ordered unwound by
China's government in April 2026 — and the company re-emerged
independent, re-raised, and competing directly with its former
acquirer. The app-layer US/China battleground — the fight is over
agents that act, not the models.
Watch: Cue's traction vs Muse/Dots; US wire pickup of the unwind
saga (none in our feeds yet); the cofounder-mobility claim;
whether "downloaded the thinking" gets corroboration (what Meta
actually retained); the compliance puzzle for enterprise buyers
(Singapore HQ, China-side ownership) and whether Western
deployments of Cue face the GLM-style scrutiny the cyber-advisory
applied to open Chinese models.