Midjourney Is No Longer Just an Image Generator
For years, Midjourney lived inside Discord. You typed /imagine, waited a few seconds, and out came something ethereal — a cat in a space helmet, a Renaissance-style portrait of your dog, a city made of candy. It was simple, beautiful, and limited to one thing: generating images from text prompts.
But 2026 has rewritten that script entirely. In the span of a few months, Midjourney has acquired a popular astrology app (Co-Star), invested over $74 million into a whole-body ultrasound medical scanner, hired architects for a luxury spa, and pushed back against Hollywood in a high-stakes copyright battle. The company that once fit inside a single Discord channel is now a multi-vertical player.
So how does Midjourney's expansion strategy stack up against the moves of AI's heavyweights — OpenAI, Google, and Meta? Let's break it down.
Product Diversification: Breadth vs. Depth
Every major AI lab is diversifying, but they're doing it very differently:
- Midjourney is going wide — apps, healthcare, wellness, legal. It's buying small teams (Co-Star's 24 employees) and experimenting across unrelated verticals. Think of it as the "startup mode" of expansion: move fast, learn cheap, pivot when needed.
- OpenAI is going deep — from GPT to Sora to o3 reasoning models, all within the AI stack. Its bet is that AI itself is the product category big enough to dominate. No spas, no astrology — just better, more capable models and API services.
- Google DeepMind is weaving AI into everything it already owns — Search, YouTube, Gemini, Workspace, Cloud, Android. Its diversification is internal: making existing products smarter rather than buying new ones in new categories.
- Meta is building the infrastructure layer — open-weight models, AI-powered social features, AR glasses, and massive compute clusters. Its bet is horizontal: give away the models, control the platform.
Healthcare: The Boldest Bet on the Board
Let's zoom in on the single most surprising move: Midjourney's $74 million investment in a whole-body ultrasound system that delivers results in under 60 seconds. This is not an obvious AI pivot.
Google has DeepMind's AlphaFold for protein folding and health AI at Verily. OpenAI has no healthcare product. Meta has none. Midjourney — the company that made AI art — is now trying to scan your organs in under a minute.
Is this genius or distraction? On one hand, medical imaging is a natural fit for computer vision AI. Midjourney's image-generation expertise could genuinely transfer to diagnostic imaging. On the other hand, healthcare is heavily regulated, capital-intensive, and slow to adopt — a sharp contrast to the fast-moving consumer AI market.
Consumer Apps: The Co-Star Gamble
The Co-Star acquisition ($4.3 million MAU, astrology app) has raised eyebrows across the industry, but it reveals a clear strategy: Midjourney wants a direct consumer app outside Discord. Co-Star's team of two dozen engineers and designers brings exactly the kind of consumer-product DNA Midjourney lacks.
- What Midjourney gains: A mobile app with an engaged user base, real-time social features, and a team that knows how to build consumer-grade products.
- What OpenAI's doing instead: ChatGPT is already a standalone app with 400M+ MAU. No acquisition needed — they grew their own.
- Google's play: Gemini is baked into Android, Google Home, and Workspace. Distribution is the moat, not the product.
- Meta's angle: AI-powered creator tools, chatbots, and smart glasses. Meta owns the social layer; AI just makes it stickier.
Approach to Openness
- Midjourney: Closed-source, subscription-only ($10–$60/month). No API access. No model weights. A pure product play.
- OpenAI: Semi-open — API-first, paid tiers, but no open-weight releases since GPT-2. Increasingly enterprise-focused.
- Google: Gemini is closed-source, but Gemma models are open-weight. Google plays both sides depending on the audience.
- Meta: The most open of the four. Llama models are freely available, and Meta actively courts the open-source community. Their revenue is ad-driven, not model-driven.
Verdict: Different Games, Different Goals
Comparing Midjourney's expansion to Big AI isn't about declaring a winner — it's about recognizing fundamentally different strategies. OpenAI and Google are playing chess on a board of 64 squares (AI models and infrastructure). Meta is building the board itself (platform and hardware). Midjourney is playing five different games at once and hoping one of them pays off big.
The question isn't whether Midjourney's strategy is "better" than OpenAI's or Google's. The question is whether a company with roughly 100 employees and limited funding can execute across apps, healthcare, wellness, and legal — all while keeping its core image-generation product competitive in a market that's getting faster and cheaper by the week.
If even one of these bets lands — the medical scanner, the consumer app, the spa experience — Midjourney could redefine what an "AI company" looks like. If none of them do, it'll still be the best image generator on the market. That's not a bad fallback plan.
Comments