Meta Cash Flow Craters as Zuckerberg Doubles Down on AI Spending

Technology · Markets

Meta cash flow collapsed by 91 percent in the second quarter of 2026 as Mark Zuckerberg pushed capital spending toward $145 billion annually, a wager that redraws the global map of digital power and leaves African governments with stark choices about infrastructure dependence.

The numbers behind the shock

Meta reported free cash flow of just $784 million for the three months ending June 2026. That figure stood at $8.55 billion in the same quarter a year earlier.

Operating cash flow reached $31.86 billion, but capital expenditure consumed $31.08 billion of it. Revenue climbed 28 percent to $60.8 billion, yet diluted earnings per share fell to $6.18 from $7.14.

The company also absorbed roughly $2.4 billion in legal charges and about $1.2 billion in severance costs from a May 2026 layoff round. Shares dropped as much as 10 percent in after-hours trading.

A capex trajectory without modern precedent

Meta’s capital expenditure guidance for 2026 now sits at $130–145 billion, narrowed from an earlier range of $125–145 billion. Total expenses for the year are expected to reach $165–169 billion.

The escalation has been steep. In 2024 Meta spent roughly $37 billion on capex; in 2025 the figure jumped to around $70–72 billion.

The 2026 plan nearly doubles that again.

Zuckerberg frames the outlay as essential for building an AI-first company. The spending covers data centres, custom silicon, and the compute clusters needed to train next-generation models.

Where the money goes: data centres as strategic assets

Meta expects to spend more than $600 billion in United States AI infrastructure through 2028, according to company forecasts. One flagship project is a $14 billion data centre venture with BlackRock in El Paso, Texas.

The company also acquired a 49 percent non-voting stake in Scale AI, a data-labelling firm, for roughly $14.3 billion in mid-2025. That deal gives Meta privileged access to the training data pipelines that underpin large language models.

These physical assets do more than serve advertisements. They concentrate compute capacity under American corporate control, shaping who can train frontier models and on what terms.

The trillion-dollar AI arms race

Meta is not alone. Alphabet, Microsoft, Amazon, and Meta are together expected to invest roughly $650–700 billion in AI-related initiatives in 2026.

Some analysts project the figure could surpass $1 trillion in 2027.

Alphabet itself recorded its first cash burn in the second quarter, as soaring infrastructure costs strained even one of the world’s most profitable firms. The pattern is consistent: cash-rich platforms are converting advertising profits into AI capacity at extraordinary speed.

For global capital markets this means trillions in savings, channelled through pension funds and sovereign wealth vehicles, are flowing into US AI infrastructure equities and bonds. America’s role as the central node of the digital economy is hardening.

What Meta cash flow stress means for Africa and the Global South

The scale of Meta’s spending carries direct consequences for African governments and firms. The company’s annual capex of $130–145 billion exceeds the entire GDP of most African nations, creating a profound negotiating asymmetry.

African regulators seeking to enforce data protection, content moderation, or taxation rules face a platform whose infrastructure investment alone dwarfs their national budgets. The ability to withdraw services or offer preferential AI access becomes a powerful lever.

Data sovereignty is also at stake. African social, economic, and health data used to train models is overwhelmingly stored and processed in US data centres under American legal jurisdiction. Local AI start-ups must typically build on US platforms, ceding margins and strategic control. These dynamics sit at the heart of the broader contest covered in our pillar, Africa: The New Scramble.

The great-power dimension

Meta’s infrastructure build is not merely commercial. It aligns with US industrial policy aimed at onshoring strategic technologies and out-competing China in artificial intelligence.

Washington has restricted exports of advanced AI chips to China, seeking to maintain a compute advantage for Western firms. Meta’s $600 billion domestic commitment reinforces that advantage physically, embedding it in concrete and fibre across American soil.

For Africa this creates a triangular competition. US platforms, Chinese cloud providers and telecom vendors, and nascent regional initiatives all vie for influence over digital infrastructure.

The choices governments make today about whose data centres and AI standards to adopt will lock in technological spheres of influence for decades.

The revenue question investors are asking

Meta’s core advertising business remains strong. Revenue rose 28 percent in the quarter, with AI-driven ad tools lifting both prices and impressions.

But the new AI products meant to justify the spending are still largely prospective.

Zuckerberg points to AI-powered smart glasses, enterprise AI services, and next-generation recommendation systems as future revenue engines. For now, capex and operating expenses are rising far faster than visible incremental income from these bets.

The market is pricing in a multi-year investment cycle with uncertain timing on returns. As long as advertising cash flow holds, Meta can sustain the spend.

Any advertising downturn would expose the strategy brutally.

What to watch next

The third quarter will test whether Meta can maintain revenue momentum while capex stays near $31 billion per quarter. Any downward revision to advertising forecasts would amplify pressure on the stock.

African policymakers should watch for signals about where Meta places its next wave of data centres. The current build is overwhelmingly American, but latency and data-localisation demands may eventually push infrastructure closer to large user bases in Africa and Asia.

The broader Big Tech capex cycle also bears monitoring. If collective spending approaches $1 trillion in 2027 as projected, the concentration of AI capacity in a handful of US firms will become a defining feature of the global economic order.

Frequently Asked Questions

Why did Meta’s free cash flow drop so sharply?

Meta’s capital expenditure reached $31.08 billion in the second quarter of 2026, nearly matching its operating cash flow of $31.86 billion. Combined with legal charges and severance costs, this left free cash flow at just $784 million, down 91 percent from a year earlier.

How much is Meta spending on AI infrastructure overall?

Meta has guided capital expenditure of $130–145 billion for 2026 and has committed more than $600 billion in United States AI infrastructure through 2028. This covers data centres, servers, and the compute clusters needed to train advanced AI models.

What does Meta’s AI spending mean for African countries?

Meta’s annual capex exceeds the GDP of most African nations, creating a sharp negotiating imbalance on data protection and taxation. African data used to train AI models is largely processed in US data centres, raising sovereignty concerns and making it harder for local AI industries to develop independently.