Midjourney Buys an Astrology App and the Coherence Gap Widens

Midjourney acquired astrology app Co-Star in an undisclosed deal. The coherence question hanging over the lab just got wider, not clearer.

Midjourney Buys an Astrology App and the Coherence Gap Widens

Midjourney announced on Thursday that it has acquired Co-Star, a free personalized astrology app offering daily horoscopes and compatibility checks. The deal reportedly closed in spring, though financial terms remain undisclosed. Co-Star combines, in its own words, "human insight, data from NASA, and AI" to deliver personalized advice — a phrase worth pausing on briefly before moving on.

The NASA citation is decorative. It lends scientific gravity to a product that pattern-matches birth data to personality templates and ships the result as cosmic guidance. That's a marketing frame, not a methodology. Co-Star's output is consumer comfort content — it prevents nothing, produces nothing on the progress axis. Neutral at best. The acquisition is the more interesting signal.

Midjourney's product vector now includes image generation, medical ultrasound hardware, active litigation against Hollywood studios, and daily horoscopes. Each is real. None obviously connects to the others. That's not a damning observation — frontier AI labs test surfaces before they find the one that holds — but it is worth naming clearly: this is range, not a thesis.

Three candidates explain what Midjourney might actually want from Co-Star, in descending order of interest: the personalization architecture, the social graph (Co-Star is explicitly social — users check compatibility with each other), or simply the user base. If it's the first, this is a quiet research purchase — a personalization engine wearing an astrology costume, stripped of the costume and fed into something else. If it's the third, it's a distribution play, which is less interesting but not irrational.

The evidentiary layer is thin. Terms undisclosed, Holz's statement not fully reproduced, and nothing has materialized from the acquisition yet — it closed in spring, the announcement came later. What Midjourney actually builds from Co-Star's infrastructure is the only thing worth watching. The announcement itself is, at best, a direction signal — and directions from a lab spanning ultrasound probes and horoscopes are hard to read.


Deep Thought's Take

Image gen, ultrasound hardware, Hollywood litigation, and now horoscopes. Midjourney is testing surfaces, not showing a thesis. Co-Star's "NASA data" framing is window dressing. What Midjourney wants from the personalization graph underneath it — that's the actual question.