Meta Releases Muse Image and Muse Spark 1.1 as AI Strategy Comes Into Focus

Meta made two significant AI announcements this week that signal a more concrete direction for its growing AI strategy — and Wall Street responded with the company's best stock week since early 2024.

Three months after launching Muse Spark, its first proprietary foundation AI model, Meta released Muse Image (a model for AI image creation targeting creators and advertisers) and Muse Spark 1.1 (aimed at agentic and coding workloads). The releases are part of a broader effort to compete directly with OpenAI, Anthropic, and Google across the AI model landscape, while also building new revenue streams beyond advertising.

Key Takeaways

• Meta's Muse Spark model family now covers image generation for its advertising and creator ecosystem (Muse Image) and agentic workloads (Muse Spark 1.1) — with Meta Superintelligence Labs, led by Alexandr Wang, driving the research agenda
• Meta expects to begin manufacturing its first custom AI chip, code-named Iris, in September — a move Bank of America analysts say may cut Meta's computing cost per megawatt significantly below Street expectations
• Wall Street's confidence shift reflects a clearer revenue roadmap: subscriptions, AI model API fees for external use, and potential cloud computing services competing with AWS and Azure — alongside its core ad business; 2026 capex guidance stands at $125–145B and may rise to $135–155B

The key shift in investor sentiment is that Meta is now explaining how its infrastructure spending connects to new revenue streams rather than just accumulating capability. For marketers and AI strategists, the Muse model family signals a direct play for the creator and advertiser ecosystem — with AI models embedded into Meta's platforms as first-party tools.

Read the full article on CNBC