Meta has partnered with Midjourney, a leader in AI image and video generation.
Meta’s Chief AI Officer Alexandr Wang announced in a Threads post on the 22nd (local time) that “Meta will collaborate with Midjourney through a technology licensing agreement to develop next-generation AI models and products.” The core of this collaboration is that Meta’s research team and Midjourney will jointly develop new AI models and products.
In the current generative AI market, OpenAI’s video generation model ‘Sora’, Google’s ‘Veo’, and Blackforest Lab’s ‘Flux’ hold the leading positions. Meta has introduced its image generation tool ‘Imagine’ and video generator ‘Movie Gen’ on its platform, but user feedback has been limited. This partnership appears to be an effort by Meta to close the gap in the AI competition and regain leadership.
Meta has recently been aggressively investing and acquiring talent. CEO Mark Zuckerberg has offered compensation packages of up to $1 billion to some researchers and invested $14 billion in data labeling company Scale AI. It has also acquired voice AI startup Play AI, and there is speculation about its potential involvement in OpenAI acquisition talks. The collaboration with Midjourney is an extension of Meta’s strategy to develop internally while also pursuing external partnerships to maintain competitiveness.
Founded by David Holz in 2022, Midjourney has rapidly established itself as a leader in the AI image generation industry within two years. Known for its realistic yet unique image style, Midjourney is projected to achieve annual revenues of $200 million by 2023. It operates independently without external investment and generates revenue through various subscription services ranging from $10 to $120 per month.
In June, Midjourney expanded its domain by launching its first video generation model, ‘V1’. Holz, the CEO, stated, “Our goal is to provide creative tools to billions, and collaboration with Meta will allow us to reach more people.”
However, this partnership brings with it the significant challenge of copyright disputes. In June, Disney and Universal sued Midjourney, alleging unauthorized use of their copyrighted materials as training data.
Meta is also not exempt from similar accusations. While recent court rulings on AI training data and copyright disputes have favored technology companies, resistance from creators and copyright holders continues to be strong.
Given this scenario, Meta’s collaboration with Midjourney amid existing copyright controversies appears to be a decision prioritizing technological advantages. Industry observers suggest that resolving copyright issues is crucial for this collaboration to succeed.
If Midjourney’s technology is fully integrated into Meta’s services, user experiences could undergo substantial changes. Features like generating custom profile pictures, adding new visual effects to posts, converting photos into specific artistic styles, and automatically creating short videos become possible. This development could enhance the competitive edge of Instagram and Facebook, where image-centric communication is vital.
Experts believe this collaboration could not only address Meta’s technical weaknesses but also set industry standards. Nonetheless, copyright concerns and safeguarding creator rights remain challenges to be addressed.
This collaboration has the potential to serve as a testing ground that goes beyond a mere technology partnership, potentially shaping the future of AI ethics and the creative environment.