Meta in double crisis: Llama 4 disappoints and 20% of staff may lose their jobs

Meta in trouble: Llama 4 disappointed and layoffs loom

It was a bad weekend for Meta AI. Reuters reports that the company is planning mass layoffs that could affect 20% of staff — over 40,000 employees — while the new Llama 4 model failed benchmark tests.

Double pressure

Meta finds itself in a bind:

• Llama 4 underdelivered — the model that was supposed to challenge GPT-5 and Claude Opus failed to meet expected benchmarks, and launch is delayed
• AI costs are exploding — Meta plans to spend nearly $135 billion in 2026, primarily on data centers and AI talent
• Revenue growth isn't keeping up — to finance AI ambitions, cuts are being made in other parts of the organization

20 percent of staff

Three Reuters sources say the layoffs could hit broadly across divisions. Zuckerberg has been clear: he wants a "year of efficiency 2.0" where AI replaces human roles internally. Now it appears that pressure from shareholders is forcing his hand.

What happens to open source Llama?

Meta has positioned itself on open source AI. But if costs become too high and revenues fail to materialize, we could see a strategy shift toward more closed or commercial Llama. This would change the playing field for all businesses currently building on free Llama models.

My take:

For CIOs who have made strategic bets on Meta/Llama as an open platform: this is a reminder that "free open source" from a publicly traded company is always conditional. Have a backup plan. And look at what's happening to Meta's own employees — 20% cuts at a tech giant isn't just Wall Street noise, it's a signal that AI is actually replacing internal roles faster than publicly admitted.