Over the past 12 months, Meta Platforms has delivered a total shareholder return of just 4%, placing it in the bottom quartile of the large-cap internet sector. This weak performance has fueled investor skepticism about the company’s ability to monetize its AI research at scale, particularly given its lack of a dedicated cloud infrastructure business. Shareholders have expressed concerns that heavy investment in proprietary data centers and custom AI hardware will not translate into immediate revenue growth, heightening scrutiny of management’s capital allocation priorities.

Meta’s capital expenditures surged to $60 billion in the fiscal year 2025, representing roughly 30% of total revenue as the company expanded its data-center footprint in Virginia, Texas and Denmark. While management expects near-term free cash flow to compress by approximately 15% in 2026, they have locked in long-term power purchase agreements totaling 2.5 gigawatts across North America and Europe at fixed rates—providing cost certainty for running AI workloads. The CFO has reiterated guidance for free cash flow conversion to recover to 40% of net income by 2028, as depreciation and build-out costs normalize.

Meta’s AI-driven recommendation engine has powered Reels to account for over 50% of all Instagram ad impressions in the full year 2025, up from just 30% in 2023. This transition has attracted a more engaged audience—average session length on Reels ads increased 25% year-over-year—and enabled a 12% rise in average price per ad impression across Facebook and Instagram. The combined effect has driven advertising revenue growth north of 20% in each of the past four quarters, underscoring Meta’s ability to reinforce its core digital-advertising business even as new initiatives scale.

Meta employs over 8,000 full-time machine-learning engineers and researchers across its AI labs in Menlo Park, Cambridge and Tel Aviv, a workforce second only to Alphabet’s research division. The company’s internal AI frameworks—Open LLM and Llama—have been adopted by more than 1,200 enterprise customers for use cases ranging from customer-service chatbots to real-time content moderation. Strategic partnerships with hardware suppliers have secured preferential access to next-generation AI accelerators, positioning Meta for dominance in both consumer-facing applications and enterprise-grade AI services over the next decade.