Meta just rolled out Muse Image, its first dedicated AI image-generation model, across the Meta AI app, Instagram, and WhatsApp. Built entirely by the company’s Superintelligence Labs division, the tool represents Meta’s clearest signal yet that it intends to own the full AI stack powering its platforms, rather than leaning on outside providers.

The market noticed. Meta’s stock climbed over 3% intraday following the announcement, hitting a one-month high. For a company spending somewhere between $115 billion and $135 billion on AI infrastructure in 2026, investors apparently liked the idea of seeing that money turn into a shipping product.

What Muse Image actually does

Muse Image isn’t just another text-to-image generator. According to Meta, the model is “agentic,” meaning it works in tandem with Muse Spark, the company’s large language model that launched back on April 8, 2026. Before generating an image, the system reasons through your prompt, searches the web for context, and plans its approach.

The tool accepts complex prompts, photo inputs, sketches, and annotations for edits. It can also blend multiple photos together. On Instagram specifically, the rollout includes over 30 effects for Stories, giving creators a new toybox of AI-powered visual tools.

Facebook and Messenger integration is coming next, which would extend the model’s reach across essentially all of Meta’s consumer-facing products.

The Superintelligence Labs strategy

Muse Image is the first model to ship from Meta Superintelligence Labs, the division led by Alexandr Wang. Wang joined Meta as Chief AI Officer in June 2025, following the company’s $14.3 billion acquisition of Scale AI, the data labeling and AI infrastructure company he founded.

Muse Image is part of the broader Muse family of models, which are gradually replacing Meta’s Llama lineup for consumer-facing applications. Muse Spark handled language. Muse Image handles visuals. A Muse Video model is reportedly in the pipeline.

This matters because Meta previously relied on external tools from companies like Midjourney and Black Forest Labs for some of its AI image capabilities. Building in-house means better integration, lower long-term costs, and full control over the training data and outputs.

The money angle

Meta’s approach appears to be a two-pronged revenue strategy. First, new subscription tiers for power users and creators. Second, and probably more important from a revenue perspective, is advertising. Meta plans to integrate Muse Image into its Advantage+ advertising campaigns, letting brands generate custom ad creative using AI.

The $115-135 billion AI infrastructure budget for 2026 is staggering by any measure. Meta needs products like Muse Image to justify that spending to shareholders, and the 3% stock pop suggests Wall Street is at least conditionally buying the thesis.