Midjourney acquires Co-Star and ventures into AI astrology

Midjourney moves beyond the creative sphere alone with the acquisition of Co-Star

Midjourney, known for its artificial intelligence image-generation tools and, more recently, video tools, has acquired Co-Star, an astrology app popular with the general public. The news was reported by The Verge, under the explicit headline “Midjourney bought the astrology app Co-Star”. The financial terms of the deal were not made public in the reported information.

At first glance, the tie-up may seem incongruous. Midjourney is one of the names most closely identified with the wave of visual generative AI that emerged after 2022. Co-Star, for its part, became known for horoscopes and astrological interpretations presented in a mobile experience with a direct, often conversational and deliberately divisive tone. Yet behind the pairing of an image generator and an astrology app, the deal reveals a broader industrial logic: generative AI is no longer only about the raw quality of a model, but about its ability to fit into a daily, recognizable product capable of retaining an audience.

Co-Star brings precisely what was not at the heart of Midjourney's historical offering: an app aimed at a non-specialist audience, a brand already established in digital cultural habits, and a personalization mechanism built around data provided by users. An astrology app does not rely on a technical demonstration comparable to image or video generation. Its value to the user lies instead in repeat consultations, editorial tone, notifications, interactions and the feeling of receiving content suited to their situation.

This type of experience is increasingly of interest to AI companies. Generative models produce text, images, audio or video on demand; they still need interfaces, habits and reasons for users to return. Major conversational assistants partly meet this ambition, but they remain general-purpose. An app such as Co-Star, by contrast, is built around a narrow and immediately understandable promise: providing a personalized reading of an astrological chart, often in a short and frequent format.

It is important to distinguish astrology's cultural dimension from scientific validity. Astrology is not an established scientific discipline and does not constitute a proven predictive tool. That clarification does not prevent its success as a cultural, conversational or entertainment practice, particularly on social platforms and in mobile ecosystems. For Midjourney, the apparent issue is therefore not to validate astrology, but to acquire a product relationship with an audience already accustomed to turning to an app for personal content.

The acquisition comes at a time when Midjourney can no longer be defined solely as an image-creation tool. The company built its reputation through a visual-generation model capable of producing illustrations, stylized images and photorealistic compositions from text prompts. It then expanded its activity into generative video. This trajectory places Midjourney in a sector where the boundaries between creative tool, content platform and consumer app are becoming increasingly blurred.

From generative imagery to a personalized relationship with the user

Midjourney's initial success was largely built among creatives, designers, online communities and users curious about the possibilities of generative imagery. Its historical Discord interface helped make creation a collective, visible activity based on experimentation. This approach had a significant effect: it enabled the company to become a reference in image generation without following exactly the path of traditional creative software.

But a creation platform does not necessarily have the same characteristics as a mass-market app. Producing an image requires an intention, a prompt, sometimes successive attempts and a minimum understanding of what the tool can do. Even when the interface becomes simpler, use often remains project-based: creating an illustration, visualizing an idea, preparing material, testing an aesthetic or making content to publish.

Co-Star follows a different logic. The user does not necessarily come to make something. They come to consult an interpretation addressed to them. The difference is fundamental for a software publisher. In a creative tool, the user is primarily a producer: they formulate a request and assess the result. In a personal-content app, they may primarily be a recipient: they receive a message, identify with it or challenge it, share it, then return later.

This distinction explains why the acquisition should not be reduced to a curiosity related to astrology. Co-Star represents an interface in which text, personalization and conversation occupy a central place. If Midjourney intends to deploy its technologies beyond visual production, an app of this type can serve as a testing ground for thinking about more intimate, more frequent products that are less dependent on explicit creative skill.

The Verge presents the deal as Midjourney's purchase of Co-Star. The important fact is less the addition of an app category to an AI player's catalogue than the change in the nature of the point of contact. Midjourney already has a strong brand in the world of generative creation. Co-Star brings a different relationship to software: that of an app that builds an editorial identity and a routine around personalized content.

This development is part of a broader trend. Since OpenAI publicly launched ChatGPT in November 2022, conversational interfaces have become one of the main gateways to AI for the general public. However, the phenomenon is not limited to general-purpose assistants. Users now encounter AI in image editors, office tools, translation services, search platforms, entertainment apps and personal-support products.

In this landscape, conversation is less an isolated feature than an interface format. It makes it possible to transform a technical model into an ongoing relationship. A tool can answer a question, suggest a rephrasing, comment on content or guide an action. Specialized apps add context to this: they know why the user is there, what kind of answer they expect and what tone is acceptable in that environment.

Co-Star has built its identity precisely around language that stands apart from the cautious or institutional phrasing of many apps. This brand voice is an asset that is difficult to reduce to a language model. It encompasses editorial choices, a product culture, references shared with its community and an ability to provoke a reaction. For Midjourney, the potential interest of such an asset is in understanding how a generative technology can fit into an experience that does not resemble a productivity dashboard.

The transition from tool to experience is not automatic. Simply integrating a language or image-generation model into an existing app is not enough to create lasting use. The experience must retain coherence, avoid giving the impression that every interaction is produced by an opaque system, and meet specific expectations. In the case of astrology, tone, trust and the relationship to privacy are particularly sensitive, because the product presents itself as personal, even when it is entertainment.

Co-Star, an acquisition that prioritizes distribution and brand

Acquisitions in artificial intelligence are often viewed through the prism of models, chips, research talent or datasets. Midjourney's acquisition of Co-Star points to another category of assets: consumer distribution. An already well-known app provides a presence in users' habits, an identity and a base of relationships with its users. In a market where many players can offer similar generative features, direct access to the public becomes strategic.

Midjourney is not the only player facing this issue. OpenAI has made ChatGPT a leading consumer interface while establishing integrations with existing products. Google is rolling out Gemini across several of its services. Microsoft has deployed Copilot throughout its software and services ecosystem. Adobe has integrated Firefly into creative tools already known to professionals and businesses. Meta, for its part, has exceptional distribution through its social and messaging apps.

These comparisons must be handled with caution. The groups mentioned do not have the same size, business model or public ambitions as Midjourney. They nevertheless show a constant: model performance is not enough to create a lasting position. There must also be an access channel, a familiar interface, a community or a product in which users already have established habits.

In this context, Co-Star represents more than just a themed app. It is a brand whose subject is immediately legible. Astrology has a strong social dimension: people discuss their sign, share interpretations, compare profiles and sometimes use this vocabulary as a light way of talking about themselves or others. An app established in this imaginary world can generate interactions that do not depend solely on the factual accuracy of an answer.

This characteristic is of interest to an AI company, but it also has a limitation. Personalized conversational products quickly face the risk of overly assertive responses, awkward wording or an impression of artificial closeness. In an app that talks about relationships, mood or personality, word choice is not secondary. It can affect the user's perception far more than in a tool that generates an illustrated background.

The acquisition therefore highlights a central market tension: companies want to make AI more personal, but greater personalization requires greater product responsibility. The boundary between a playful experience, wellness advice and an interaction that may influence behavior must remain understandable. The teams responsible for design, moderation and support play a decisive role in defining that boundary.

The Co-Star case also serves as a reminder that a consumer brand cannot be reduced to its technology. Value may lie in the mobile experience, writing, content cadence, notification mechanisms and the way a community takes ownership of the product. For Midjourney, which experienced much of its growth through a creator community, this aspect is not unfamiliar. But Co-Star's audience does not primarily come looking for settings, visual styles or technical mastery. It expects a concise, accessible experience centered on itself.

The acquisition can thus be read as a shift in focus: from demonstrating a model's capabilities to designing a service. This nuance matters. Competition in generative AI is often described as a race for the next model version. Yet as technologies spread, differentiation also depends on the ability to design products whose usefulness, tone and frequency of use are distinctive.

Trust, data and compliance issues in Europe

For French and European users, the tie-up between Midjourney and Co-Star immediately raises the question of personal data. An astrology app may request information such as date, time and place of birth in order to produce a chart or interpretation. It may also process interactions carried out within the app. These elements are not necessarily all sensitive data in the strict sense of European law, but they contribute to building a personal profile and may reveal habits or interests.

The General Data Protection Regulation, the GDPR, governs the processing of personal data in the European Union. For an app that aims to be personalized, the principles of transparency, data minimization, purpose limitation and security are particularly important. Users must be able to understand what data is collected, why it is collected, how long it is retained and what choices are offered to them.

The arrival of a generative AI player in the equation may increase questions, even without presuming the technical or commercial choices that will follow the acquisition. If generative systems are involved in creating messages, personalization or conversational interactions, the quality of information provided to the user becomes central. A personalized experience must not become a black box in which data is collected without the service's general operation being intelligible.

The European framework on artificial intelligence, the AI Act, also reinforces the role of transparency and risk-management obligations in the debate. Its application is gradual and depends on the categories of systems concerned. It would not be accurate to infer, without specific information, which obligations would apply to each component of any potential product resulting from the tie-up between Midjourney and Co-Star. However, the European regulatory trend clearly points toward greater requirements for providers and deployers of AI systems.

For a company targeting the general public, compliance is not only a legal matter. It is also a reputational issue. The European market pays close attention to data protection, explanations of processing and the safety of minors. Astrology attracts diverse audiences, including young people. An app using conversational mechanisms must ensure that it does not present itself as a medical, psychological, financial or legal authority.

The difference between entertainment content and personalized advice can become blurred when an interface adopts a highly intimate voice. This is one of the challenges of all conversational AI, whether it addresses astrology, relationships, education or well-being. The more targeted a response appears, the more the user may attribute an intention or expertise to it that it does not possess. Editorial safeguards, the wording used and reporting mechanisms then take on concrete importance.

France is a particular setting for this discussion. The country has an active technology and creative ecosystem, but also a strong culture of data protection, notably driven by public attention to the positions of the National Commission on Informatics and Liberty. Generative tools are already used by freelancers, agencies, studios, educational institutions and businesses. However, moving from these tools to highly personal conversational products changes the nature of the questions raised.

In the French-speaking creative world, Midjourney is associated with both professional uses and amateur practices: graphic exploration, design, mood boards, illustration and prototyping. Co-Star introduces another logic of consumption, more closely tied to daily life and identity. If Midjourney manages to bring these two worlds together without dissolving the specific features of each, the company could broaden its understanding of consumer uses. But this expansion also requires greater discipline in designing services likely to collect personal information and produce emotionally charged messages.

Lastly, it should be remembered that an acquisition alone does not indicate how products will be integrated. The Verge reports that Midjourney bought Co-Star, but this does not make it possible to state that the app will adopt a particular feature, that its data will be used in a particular way, or that image and video generation will be added to it. Between the announcement of a deal and the concrete evolution of a product, many scenarios remain possible: maintaining the app, bringing teams closer together, changing infrastructure, experimenting with new interfaces or retaining broad editorial autonomy.

A revealing diversification in the new economy of AI products

The strategic interest of the acquisition can be understood through a simple question: how does an AI company turn a technical capability into a lasting business? Generative models can be impressive, but their development, inference and update costs require finding regular uses. Subscriptions are one possible answer, as are professional offerings, licensing or integrations into existing platforms. Specialized consumer apps offer another model: they create a recurring reason to return around a clear promise.

Image and video creation remains an important market, but it is highly competitive. Midjourney operates alongside offerings from major technology companies and creative software publishers. OpenAI has made its mark on the image sector with DALL-E, while Adobe has placed Firefly at the center of its generative AI strategy for creation. Google, Microsoft, Meta and numerous specialized players also have capabilities or products related to multimodal generation.

In this environment, owning a product that is not defined exclusively by a model's output can become a form of diversification. Co-Star does not compete directly with an image editor. It belongs to a category where value depends on user experience, habit and brand. For Midjourney, the deal may therefore reduce its symbolic dependence on a single question: which model produces the most convincing image at a given moment?

It would nevertheless be premature to present the acquisition as proof of a fully settled strategy. The terms of the deal were not detailed in the cited information, and Midjourney's precise intentions for Co-Star are not established by the acquisition alone. What can be observed, however, is that the company has chosen to invest in a product far removed from its usual public image. That distance is itself significant.

It suggests that the AI market is not structured solely by technical modalities — text, image, audio, video — but also by contexts of use. The same infrastructure can potentially support very different experiences. The issue is no longer simply generating content on command. It is determining when, why and in what form that content is requested, received, shared or ignored.

General-purpose assistants have shown that a conversation can become a universal interface. Specialized apps, by contrast, seek to make that conversation more situated. They do not answer every request; they promise to answer a family of needs or curiosities. This specialization can make the experience clearer, but it requires the product to remain coherent and not add AI merely to follow a trend.

For the French-speaking sector, the deal is also a signal for app publishers and startups. The race is not limited to training large models, a field dominated by companies with considerable resources. There is also a battle for products, communities, personalization and interfaces. An app with a loyal audience can become strategic for a player seeking to bring its AI capabilities closer to everyday uses.

This observation applies particularly to cultural, educational, creative or leisure apps. They often have a fine-grained understanding of their audience and its expectations. They can also be vulnerable if general-purpose platforms absorb their functions. The acquisition of Co-Star shows, conversely, that a vertical app can represent an attractive asset when it has succeeded in creating a strong identity and a repeated relationship with its users.

Toward less visible AI, but more present in digital routines

The most important prospect opened up by this deal concerns the normalization of AI in products where it is not necessarily highlighted as a technological feat. Midjourney built its reputation on immediately spectacular images: the user sees the model's result, compares it, shares it and discusses it. An app like Co-Star relies more on the continuity of an experience. If AI plays a greater role there in the future, it could be less visible as technology and more present as a mechanism for personalization, dialogue or editorial production.

This development is already perceptible in several categories of software. Generative functions are increasingly integrated into existing tools rather than presented as separate destinations. Users do not always ask for an “AI model”; they want to write faster, prepare an image, find information, rephrase a text or receive a suggestion. Value then shifts toward the quality of integration and the trust placed in the service.

The challenge for Midjourney will be not to confuse personalization with automation. A consumer app can become more compelling if it is more responsive, more creative or better suited to the context. But it can lose its identity if responses become generic, if the editorial voice becomes diluted, or if the user feels they are interacting with a system that merely seeks to maximize their attention. In an astrology product, this boundary is particularly fine because the content directly touches on how people see themselves.

Co-Star's future under Midjourney ownership will therefore depend less on an abstract AI promise than on concrete product decisions: the place given to Co-Star's team and brand, transparency around changes, data processing, consistency of the experience and the ability to preserve a clear distinction between entertainment, creative expression and advice. These decisions are not documented by the announcement reported by The Verge, and it would be risky to anticipate their details.

One thing is nonetheless clear: the acquisition broadens Midjourney's range of possibilities. By moving from an environment identified with visual creation to an app based on daily return and personalization, the company is potentially testing another way of existing in the public sphere. For European and French players, this deal is a reminder that the next AI battles will take place as much in interfaces and cultural habits as in model performance rankings.

In the long term, the best-positioned companies will not necessarily be those that present the most spectacular generative feature during a demonstration. They may be those that manage to establish this technology in repeated, understandable and sufficiently reliable uses for users to return to it. With Co-Star, Midjourney is moving in that direction through an unexpected door: not the creative studio, but the personal, conversational and cultural app.