Meta Is Spending $370 Million Per Day on AI — And Firing 8,000 People to Pay for It

The numbers are staggering. Meta has raised its 2026 AI infrastructure budget to between $125 billion and $145 billion.

That works out to approximately $370 million every single day being poured into data centre construction.
At the same time, Meta has announced 8,000 layoffs — 10% of its entire workforce — effective May 20. Recruiting and HR departments face cuts of 35 to 40%.
The message from Mark Zuckerberg's company is unmistakable: humans are being replaced by infrastructure. The workers who found and hired talent are among the first to go. The machines they were indirectly building are taking their place on the balance sheet.

Investors reacted with scepticism.

Meta's stock fell more than 6% after its Q1 earnings, despite revenue jumping 33% year-over-year. The concern is structural: unlike Google and Amazon, Meta does not have a cloud computing business to monetise its AI investments directly. It is spending at cloud-competitor scale without cloud-competitor revenue diversification.
Meta's open-source Llama models remain a genuine strategic asset — widely adopted, developer-friendly, and increasingly capable. But the question Wall Street is asking is simple: when does the spending stop, and when does the return begin?

The European perspective: Meta's mass layoffs coincide with its heaviest AI investment in history. European labour law and worker protection frameworks make this trade-off far more difficult to execute in the EU — and far more politically visible.