Meta’s New Muse AI: Zuckerberg’s Billion-Dollar Bet

Meta unveiled a new flagship artificial intelligence model called Muse Spark on April 8, 2026. The release is the first major product from Meta Superintelligence Labs, a division created following a significant restructuring of the company’s AI operations in 2025.

Muse Spark is currently available to users through the Meta AI app and the meta.ai website. Unlike the Llama series, Meta is keeping Muse Spark closed source for the time being, though the company has indicated that future iterations in the Muse lineup may be released as open source.

Shift Toward Agentic AI

The development of Muse Spark is part of a broader strategy to move beyond conventional AI assistants. Mark Zuckerberg has stated that the goal is to create AI products that don’t just answer your questions but act as agents that do things for you.

This approach aligns with a vision for personal superintelligence that Zuckerberg first detailed in a manifesto published in July 2025. According to that vision, the AI is intended to help individuals pursue their own specific goals rather than being controlled by a top-down architecture.

In terms of functionality, Muse Spark is designed for everyday personal utility. Its primary capabilities are focused on several key areas:

• Visual understanding
• Health-related tasks
• Shopping assistance
• Social content creation

Zuckerberg expressed optimism that these capabilities will support a wave of creativity, entrepreneurship, growth, and health.

Strategic and Technical Pivot

The launch of Muse Spark serves as a credibility effort for Meta following the April 2025 release of Llama 4. Llama 4 received a tepid reception from the industry and was widely viewed as underwhelming.

The transition to a closed-source model for Muse Spark marks a departure from Meta’s previous role as a leader in open-weights AI. While Llama models were made freely available to hobbyists, startups, and researchers, Muse Spark is not available for download.

Meta describes Muse Spark as the first step on our scaling ladder and the first product of a ground-up overhaul of our AI efforts.

Investment and Organizational Overhaul

The creation of Muse Spark followed an aggressive period of spending and hiring. In June 2025, Meta spent more than $14 billion to recruit Alexandr Wang, the co-founder of Scale AI, to serve as the centerpiece of the company’s AI overhaul. Wang now serves as Meta’s chief AI officer.

To support this new direction, Meta recruited more than 50 researchers from competing firms, including Google, Anthropic, and OpenAI. This hiring surge led to the formation of Meta Superintelligence Labs, which was later restructured into four specialized units:

• Research
• Superintelligence development
• Products
• Infrastructure

The financial commitment to these efforts is substantial. In January 2026, Meta informed Wall Street of plans to invest between $115 billion and $135 billion into capital expenditures for the year 2026, nearly doubling the company’s 2025 capex figure.

This level of spending has drawn scrutiny from analysts. Morningstar analyst Malik Ahmed Khan noted that after a year of aggressive hiring and limited releases, Meta needed to demonstrate to operators and investors that it had been working on something of substance.

I think Meta had to show investors and operators they have been working on something of substance. That’s the first step.

Malik Ahmed Khan, Morningstar analyst

As Meta shifts toward proprietary models, the company faces the challenge of establishing new revenue streams to justify its massive investments in AI infrastructure and talent.