Meta Just Proved That AI Spending Without AI Strategy Is Just Expensive Firing
The $135 Billion Question Nobody Is Asking
Meta cut 700 jobs this week across Reality Labs, Facebook, recruiting, and sales. That brings the 2026 total to roughly 2,200 positions eliminated — on top of 3,600 "performance-based" terminations in early 2025.
The real story isn't the layoffs. It's what Meta is doing with the money.
The company's 2026 capital expenditure forecast sits at $115–135 billion, nearly double 2025's $72 billion. Total projected expenses: $162–169 billion. Barclays forecasts a near-90% drop in free cash flow. Meta is betting the entire company on AI infrastructure — data centers, Nvidia GPUs, custom chips, and its Llama ecosystem.
Most coverage frames this as "Meta invests in AI." That framing is dangerously incomplete.
Why Spending More on AI Makes Most Companies Worse
Here's what the LinkedIn consensus gets wrong: spending on AI infrastructure is not the same as having an AI strategy. Meta's own numbers prove it.
Reality Labs has lost $90 billion cumulatively since its creation. In 2025 alone, the division burned $19.2 billion. Meta's response? Close three VR studios (Twisted Pixel, Sanzaru Games, Armature Studio), cut 10% of Reality Labs in January, and pivot from VR headsets to smart glasses and wearable AI.
That's not strategy. That's expensive trial-and-error with a $135 billion budget.
The pattern is clear across the industry. 45,000+ tech jobs were eliminated globally in Q1 2026, with AI cited as the driving force in 1 out of 5 cases. Atlassian cut 1,600. Amazon cut 16,000. Block cut 4,000. Every company says the same thing: "We're reallocating to AI."
But reallocating from what to what? Most can't answer that specifically. And that's the problem.
What the Top Engineering Leaders Actually Do Differently
The companies getting AI investment right share three patterns that Meta's approach violates:
| Pattern | What Winners Do | What Meta Did |
|---|---|---|
| Measure before spending | Quantify where AI creates margin, then fund those areas | Doubled capex across the board, then cut teams to afford it |
| Protect institutional knowledge | Keep senior ICs and domain experts through transitions | Cut across recruiting, sales, and product — losing tribal knowledge |
| Ship before scaling | Prove ROI on small AI deployments, then scale what works | Committed $135B to infrastructure before Llama monetization is proven |
Zuckerberg claims output per engineer increased 30% since early 2025 thanks to AI coding tools. Even if true, that metric measures velocity, not value. Shipping faster doesn't matter if you're shipping in the wrong direction — and $90 billion in Reality Labs losses suggests direction is exactly the problem.
The companies that will win the AI transition aren't the ones spending the most. They're the ones who know exactly which problems AI solves for their customers, and refuse to fund anything else.
- Stripe invested in AI fraud detection — specific, measurable, tied to revenue. Not "AI infrastructure."
- Shopify told employees that before requesting headcount, they must prove AI can't do the job first. That's constraint-driven strategy.
- Linear built AI into the core product loop, not into a separate division burning billions with no clear path to profitability.
The difference between these companies and Meta isn't budget. It's clarity.
Final Takeaway
If your AI strategy requires firing hundreds of people to fund it, you don't have an AI strategy. You have a cost problem dressed up in GPU purchase orders. The companies that win this transition will be the ones that can explain — in one sentence — what their AI investment produces for customers. Meta still can't.
Try Software Insight
Massive AI spending means nothing if you can't measure what it produces. Software Insight gives engineering leaders a clear view of delivery risk, team maturity, and quality metrics — so your investment decisions are based on data, not hype.