Meta's Desperate Pivot: Muse Spark Shows Big Tech Finally Admits It's Losing

Meta debuted Muse Spark, its first major AI model since hiring Scale AI's Alexandr Wang nine months ago, dubbed Muse Spark and originally code-named Avocado, as the first from Meta Superintelligence Labs. But here's what matters: Meta is abandoning the open-source strategy that defined Llama.

Meta said its AI-related capital expenditures in 2026 will be between $115 billion and $135 billion, or nearly twice its capex last year. Spend is doubling. Performance isn't matching it.

While Meta has used advancements in generative AI to bolster its advertising business, it's yet to crack the AI model market in a significant way, with OpenAI and Anthropic now collectively valued at over $1 trillion, and Google's Gemini technology gaining traction in the consumer market.

Muse Spark uses a squad of AI agents to help "reason in parallel," helping it "compete with the extreme reasoning modes of frontier models such as Gemini Deep Think and GPT Pro". The language itself tells the story: we're copying, not leading.

My take: This is the sound of a company realizing that open-source moats don't work in frontier AI. When you're spending $125B+ on capex, you can't afford to give away your competitive advantage. Meta's shift to Muse Spark is less about innovation and more about damage control. The real question: can $125B in spend actually buy dominance when you're 18 months behind on momentum?

Sources