Meta undercuts rivals with low-cost AI model
Meta Platforms Inc. released the Muse Spark 1.1 AI model through a public Model API, pricing it significantly lower than rivals to capture enterprise market share. The company plans to increase capital expenditures to $142 billion in 2026 and expand compute capacity to 14 gigawatts by 2027 to support this growth. JPMorgan views this as a pivotal step in Meta's strategy to monetize its substantial AI investments outside of advertising.

*this image is generated using AI for illustrative purposes only.
Meta Platforms Inc. launched its Muse Spark 1.1 artificial intelligence model through a public preview of its new Model API on Thursday, marking the company's first significant step toward monetizing its AI technology externally. CEO Mark Zuckerberg announced the release, highlighting the model's improved capabilities in agentic reasoning, coding, and tool use. The launch represents a strategic shift as Meta seeks to generate returns on its massive AI investments, with JPMorgan noting that the API pricing is set at roughly 25% of the cost of leading models from OpenAI and Anthropic.
Pricing and Market Position
Meta's aggressive pricing strategy aims to quickly gain traction with developers and enterprise customers. By offering the Muse Spark 1.1 model at a fraction of the cost of competitors, Meta intends to disrupt the enterprise AI market. JPMorgan analyst Doug Anmuth suggested that this pricing approach, combined with the model's advanced capabilities in multimodal reasoning and computer use, could help Meta narrow the competitive gap with OpenAI, Anthropic, and Alphabet Inc.'s Google.
Capital Expenditure and Infrastructure
The monetization push comes as Meta substantially increases its capital expenditure to support its AI ambitions. JPMorgan projects Meta's capital expenditures will reach $142 billion in 2026, an increase of 104% year over year, before climbing to $202 billion in 2027. To support this scale, Reuters reported that Meta plans to expand its AI compute capacity to 7 gigawatts in 2026 and 14 gigawatts in 2027. This infrastructure is intended to support both internal AI products and new external business ventures.
Internal Integration and Strategy
Meta's internal AI strategy remains a critical component of its overall approach. The company has integrated tools such as DevMate, Metamate, and Google's Gemini into its engineering workflows, with some teams setting targets for AI to assist with the majority of code changes. This dual strategy focuses on driving operating leverage by improving developer productivity and shortening product cycles. Additionally, Meta continues to expand consumer-facing AI products, including the Muse Image generator and new features on Instagram, despite some scrutiny regarding user privacy settings.
| Financial Projection | 2026 Estimate | 2027 Estimate |
|---|---|---|
| Capital Expenditure | $142 billion | $202 billion |
| AI Compute Capacity | 7 gigawatts | 14 gigawatts |
Future Outlook
JPMorgan maintains a Neutral rating on Meta with a $725 price target, citing the company's improving AI models and early monetization efforts. The firm noted that developer adoption and enterprise demand could drive upside, particularly if Meta expands beyond its core advertising business. Zuckerberg also indicated that excess computing capacity could eventually be rented out, providing another potential revenue stream if internal demand does not fully utilize the expanded infrastructure.
How will competitors like OpenAI and Anthropic respond to Meta's aggressive pricing strategy?
What are the potential revenue implications if Meta begins renting out its excess computing capacity?
Will the significant increase in capital expenditures pressure Meta's free cash flow in the short term?

































