Meta launches Muse Image AI as it builds cloud business
Meta has released its first image-generation model under Alexandr Wang's lab, a move that expands its consumer AI reach while laying the groundwork to sell cloud computing power to external developers.
Meta launched Muse Image on Tuesday, marking the first image-generation model developed under its Superintelligence Labs division. The unit is led by Chief AI Officer Alexandr Wang. The tool is rolling out immediately across the Meta AI chatbot, Instagram, and WhatsApp, allowing users to generate images from text prompts or alter existing photographs.
A standout feature allows users to create images featuring friends or creators based on their public Instagram posts. Individuals who do not want their content remixed by the AI can opt out through the settings menu. This approach relies on an opt-out mechanism rather than explicit consent, a distinction that will attract intense scrutiny from European data protection regulators.
To address misuse, all generated images carry an invisible watermark. Meta has also installed safety precautions to block material that violates its terms of service, including child sexual abuse material. Advertisers will gain access to the model soon to produce marketing materials.
Muse Image is the second major product from the lab Wang was hired to lead a year ago, following the April release of the Muse Spark large language model. To build this division, Meta invested billions and hired researchers with compensation packages exceeding $200 million.
The strategic importance of these models extends beyond Meta’s own social platforms. The company plans to sell access to its AI models to outside developers through a cloud offering. This move is designed to monetise Meta's expanding data centre infrastructure and its costly stockpile of AI chips.
According to a Bloomberg report last week, this cloud business could eventually include selling raw computing power. If realised, this would position Meta as a direct competitor to established cloud providers.
Despite these commercial ambitions, Meta remains highly dependent on external partners for computing capacity. The company recently signed deals to rent infrastructure from CoreWeave, Google, and Oracle. This aggressive spending on AI infrastructure followed a decision in May to cut 8,000 jobs and redirect billions of dollars toward the technology.
The rapid product releases underscore Meta's urgency to compete with OpenAI and Anthropic across every AI modality. A video generation model from Wang's lab is expected in the coming months, further escalating the compute arms race.
