Co-Star
2026
Why
Expand into standalone social applications.
Midjourney, Inc.
Acquisitions
1
Major Deals Covered
This page covers 1 major deal, each with its price where disclosed and the reason behind it.
| Company | Year | Price | Purpose |
|---|---|---|---|
| Co-Star | 2026 | Undisclosed | Expand into standalone social applications. |
2026
Expand into standalone social applications.
Growth comes from regular model releases (V7 in April 2025, Video V1 in June 2025, Niji 7 in January 2026, V8 alpha in March 2026, V8.2 in July 2026), a web app that removed the need for Discord, and licensing: in August 2025 Meta announced it would license Midjourney's image and video technology. Holz has said the long-term goal is real-time open-world simulation.
Midjourney sells subscriptions. Plans are Basic ($10/month), Standard ($30), Pro ($60) and Mega ($120), with discounts for annual billing. Each tier buys a monthly allotment of Fast GPU time; Standard and above add unlimited Relax-mode image generation, and Pro and Mega add Stealth Mode so creations stay out of the public gallery. Extra Fast hours can be bought on demand. There is no advertising and no free tier for ordinary image generation, so GPU compute and staff are the main costs set against subscription income. Usage runs through midjourney.com and the original Discord bot.
No acquisitions by Midjourney have been publicly reported as of October 2026. It builds its models and web app in-house.
Midjourney maintains a strict organic research philosophy, avoiding large corporate mergers or acquisitions of legacy media companies. Instead, Midjourney focuses on targeted acqui-hires of top AI researchers, spatial computing hardware engineers, and 3D graphics specialists.
Unlike Adobe, Midjourney does not acquire stock photography libraries. Instead, Midjourney trains its foundation models on large-scale open web datasets, paired with proprietary synthetic data generation, aesthetic curation, and human preference RLHF scoring.
Midjourney built its web canvas, inpainting engine, and social feed entirely in-house using lightweight, high-performance web technologies, ensuring smooth real-time synchronization with its Discord backend and distributed GPU clusters.
Rather than acquiring external video generation startups (like Runway or Pika), Midjourney's core research team in San Francisco is developing proprietary multi-modal video synthesis and real-time 3D voxel generation engines natively from scratch.
Midjourney closely tracks academic breakthroughs in latent diffusion, transformer architectures (DiT), and flow matching, integrating state-of-the-art open research principles into its closed, highly optimized proprietary production models.
Midjourney has executed zero external corporate acquisitions. Instead of pursuing traditional tech M&A, Midjourney maintains an ultra-lean team of under 100 researchers and engineers, investing capital directly into high-performance cloud compute partnerships (particularly with Google Cloud TPU infrastructure) and proprietary model architecture research.
In 2024, Midjourney recruited Ahmad Abbas, former hardware engineering manager for Apple's Vision Pro, to launch a dedicated hardware division in San Francisco. While image and video generation remain its core software engines, Midjourney is developing proprietary physical hardware devices designed to bridge generative AI with interactive spatial and physical environments.
Compare corporate acquisition records, deal values, and strategic targets
Explore Midjourney
Core profile pages, founders and related research hubs for this company.