The Massive Strategic Shift: Open Source to Proprietary

For years, Meta’s AI strategy was built on open-source models. Its Llama family gave developers free access to powerful base models — a strategy designed to build an ecosystem and prevent OpenAI and Google from monopolising AI infrastructure.

With Muse Spark, Meta has made a dramatic reversal. The new model is proprietary — closed-source, not freely downloadable, and not available for developers to run on their own hardware. This is a fundamental change that has major implications for Meta’s business.

Why the change? Two words: monetization pressure. Meta’s stock surged 10% in the week following Muse Spark’s launch — but investors have been watching Meta spend hundreds of billions on AI infrastructure while generating essentially zero direct AI revenue. That had to change.

The $14.3 Billion Investment

In June 2025, Meta made one of the largest bets in tech history: spending $14.3 billion to acquire a 49% non-voting stake in Scale AI and bringing its CEO Alexandr Wang in-house as Meta’s first-ever Chief AI Officer. Wang then built Meta Superintelligence Labs from the ground up, recruiting researchers from OpenAI, Anthropic and Google.

On top of the Scale AI investment, Meta committed $115–135 billion in capital expenditure for 2026 — nearly double the prior year — including construction of the Hyperion data centre (one of the largest AI computing facilities ever built).

Muse Spark’s Revenue Streams

Paid API Access — The Primary Bet

The clearest and most direct path to Muse Spark revenue is through a paid developer API. Meta is currently running a private API preview with select partners — and a public API is expected to launch soon.

For context, OpenAI generates billions of dollars annually from API access to GPT models. Anthropic generates significant revenue from Claude API access. If Muse Spark can attract even a fraction of the developer market, the revenue potential is enormous.

Meta’s competitive advantage for API: price and efficiency. Because Muse Spark achieves competitive performance with 10x less compute than Llama 4 Maverick, Meta can potentially undercut OpenAI and Anthropic on price while maintaining healthy margins.

AI-Powered Advertising — The Hidden Goldmine

Meta’s primary business is advertising — generating $201 billion in revenue in 2025, almost entirely from ads on Facebook, Instagram, Threads and WhatsApp. Muse Spark has the potential to significantly enhance this core business.

Muse Spark’s world-class multimodal vision capabilities are particularly valuable for advertising:

• Creating smarter, more visually compelling ad creatives from product photos
• Analysing ad performance with deeper visual understanding
• Generating personalised ad content based on user interests and behaviour
• Powering a “shopping mode” that blends AI intelligence with Meta’s user data
• Enabling dynamic video ad creation through the Vibes AI feature

One analyst quoted by CNBC noted that Muse Spark’s visual strengths make it particularly well-suited for advertisers: “Compared to Claude and Gemini, it definitely feels like it has more of a consumer bent” — designed for the visual, short-form content world that Meta’s advertising dominates.

What This Means for Developers

The shift to proprietary AI is a significant change for the developer community that relied on free Llama models. Here’s the honest picture:

• No more free self-hosting of Meta’s frontier models — Muse Spark cannot be downloaded
• API dependency — developers will need to pay Meta for access, as they do with OpenAI/Anthropic
• Open-source future unclear — Meta says it “hopes to open-source future versions” but has made no commitment
• Potentially competitive pricing — Meta’s compute efficiency could translate to lower API costs
• Deep Meta ecosystem integration — building on Muse Spark means deep integration with Meta’s social platforms

Investor Reaction

Wall Street responded enthusiastically to Muse Spark’s launch. Meta’s stock surged nearly 10% in the five trading days following the April 8, 2026 announcement, closing at $629.86 per share on Friday.

The single-day jump of 6.5% outpaced the S&P 500’s 2.5% gain and Nasdaq’s 2.8% rise on the same day. Analysts cited several reasons for the bullish reaction:

• Proof that Meta’s $14.3 billion Alexandr Wang bet was producing results
• A credible path to AI monetization through API access
• The proprietary model approach signals Meta is “in the game” as a serious AI competitor
• Muse Spark’s advertising potential enhancing Meta’s core revenue engine

Meta trades at an attractive 18x P/E multiple — below Alphabet (23x) and Microsoft (19.6x) — suggesting the market had not fully priced in Meta’s AI potential before this launch.

Can Meta Make Money: The $325 Billion AI Market Opportunity

The prize Meta is competing for is enormous. According to Grand View Research, the global generative AI market is projected to grow from $22 billion in 2025 to $325 billion by 2033 — a compound annual growth rate of over 40%.