Meta may halt $174B buyback program to fund AI infrastructure

Meta Platforms' leadership, under billionaire CEO Mark Zuckerberg, has pursued high-growth initiatives including the metaverse and artificial intelligence. But for the past decade, the company's most consistent capital allocation tool has been its share repurchase program. Beginning in 2017, Meta has systematically bought back its own stock each year: $1.976 billion（約3200億円） in 2017, $12.879 billion（約2.1兆円） in 2018, climbing to $44.537 billion（約7.1兆円） in 2021, and then moderating to $26.248 billion（約4.2兆円） in 2025. Cumulatively, these buybacks have totaled approximately $174 billion（約28兆円） and retired nearly 12.7% of the company's outstanding shares. For a company with steady or growing net income, this declining share count has translated into higher earnings per share—a powerful attraction for value-seeking investors.

But Meta's AI ambitions are forcing a hard choice. According to reports, the company has not repurchased shares since Q3 2025 and is weighing the option of issuing equity and undoing some of its share buybacks to fund its AI infrastructure build-out. The trigger is Meta's sharply increased forecast for AI-related capital expenditures. This pivot reflects the scale of the bet: the combined CapEx of Amazon, Google, Meta, and Microsoft is projected to surge 98% year-over-year to a record $715 billion（約110兆円） in 2026—nearly 3 times the 2024 amount and more than 5 times 2023 levels.

There are near-term offsets to that burden. Meta has integrated generative AI into its advertising platforms, allowing clients to tailor static and video messages to users and improve click-through rates, which strengthens Meta's ad pricing power. The company has also unveiled plans to sell excess AI data center compute capacity, a revenue stream intended to ease the impact of its massive capex. Yet the risks are real. History shows that every game-changing technology over the past three decades has experienced an early-stage bubble-bursting event—a pattern Meta shareholders know intimately from the 2022 metaverse collapse. If an AI bubble were to burst, Meta's stock would likely face near-term pressure, and without share buybacks acting as an added catalyst to support earnings per share, the stock could be weighed down more sharply.

Meta's decision to suspend share repurchases marks a strategic pivot driven by unprecedented capital demands in AI infrastructure. For nearly a decade, the company's buyback program—which accumulated $174 billion（約28兆円） and retired 12.7% of outstanding shares—has been a pillar of shareholder returns, mechanically boosting earnings per share despite flat or modest net income growth. By halting this program in Q3 2025, Meta is signaling that near-term stock support through financial engineering is taking a back seat to long-term competitive positioning in artificial intelligence.

The shift reflects genuine business opportunity: Meta is already capturing near-term AI monetization through generative AI in its advertising platform, which helps clients tailor messages and strengthens the company's ad pricing power. The company is also planning to sell excess AI data center compute capacity, a revenue stream that may partially offset the staggering infrastructure costs. However, the scale of that investment is daunting—the combined 2026 CapEx of Meta, Amazon, Google, and Microsoft is expected to reach $715 billion（約110兆円）, nearly 3 times 2024 levels. For value-oriented investors accustomed to Meta's earnings-per-share tailwind from buybacks, the loss of that mechanical support could weigh on the stock, especially if AI returns disappoint or the sector experiences a cycle correction similar to the 2022 metaverse collapse.