META | Weekly | Haru Insights

Meta announced a $31 billion capital expenditure in Q2 2026, triggering a 91% collapse in free cash flow and compressing operating margins to 31%—a bold, high-stakes bet on AI infrastructure that underscores its commitment to becoming a foundational player in the next generation of artificial intelligence.
The company finalized a $6 billion fiber optics agreement with Corning and launched Muse Spark 1.3, marking a decisive pivot toward proprietary AI infrastructure and full developer access to user prompts and outputs, cementing Meta’s bid to control the data and hardware backbone of AI innovation.

Meta’s $31 billion capex surge is not just spending—it’s a strategic transformation, repositioning the company as a central architect of AI infra, not just an app developer, with far-reaching influence over the future of machine intelligence.
Muse Spark 1.3’s aggressive pricing and full access to user prompts and outputs are creating a moat of real-world training data, giving Meta an unprecedented edge in refining AI models at scale and accelerating commercial deployment.
The $6 billion Corning fiber deal is a foundational move to bypass third-party bottlenecks, ensuring Meta can deliver ultra-low latency and massive bandwidth—critical for next-gen AI workloads—well ahead of competitors.
By internalizing AI infrastructure and data pipelines, Meta is no longer dependent on cloud giants, asserting full control over performance, security, and innovation speed in its AI ambitions.
Meta’s pivot to open developer access with Muse Spark 1.3 signals a deliberate strategy to build a thriving ecosystem, fostering innovation while capturing invaluable feedback loops that fuel continuous, rapid model iteration.

Meta Platforms, Inc. (META)[Tv]