Meta and Microsoft Cut 20,000 Jobs. Big Tech Is Spending $700B on AI. Both Are True.
By easyAI Team · 12 min read · 2026-04-25
The same companies spending $700 billion on AI this year are cutting roughly 20,000 jobs this month. Meta is eliminating 8,000 positions. Microsoft is offering voluntary buyouts to up to 8,750 people. The stock market responded positively to both announcements.
That's the version you saw in the reel. The full version requires understanding why companies pour hundreds of billions into one line item while cutting thousands from another, and what it means that Wall Street treats this as good news.
What Happened at Meta?
On April 23, 2026, Meta notified employees that approximately 8,000 people would be laid off. That's roughly 10% of the company's full-time workforce. The layoffs take effect May 20, 2026.
On top of the 8,000 cuts, Meta also closed approximately 6,000 open positions. Those are roles that were approved, budgeted, and in many cases actively being interviewed for. Combined, that's 14,000 roles that no longer exist at Meta.
The decision came from CEO Mark Zuckerberg. The stated reason: restructuring for AI investment efficiency. Meta is reallocating resources, not shrinking. The company is spending more money than ever. It's just spending it on infrastructure instead of people.
What's the Microsoft Buyout?
Microsoft's approach is structurally different from Meta's.
Microsoft offered voluntary buyouts to approximately 7% of its U.S.-based employees. With a U.S. workforce of about 125,000, that means up to 8,750 people could take the package if every eligible employee accepts.
The word "voluntary" matters. Meta's layoffs are forced. If your role is eliminated, you leave. Microsoft's buyouts are opt-in. You choose whether to take the package and go.
In practice, voluntary buyouts and forced layoffs often produce similar outcomes. Companies offer buyouts when they want to reduce headcount without the legal and PR costs of involuntary termination. Employees who take buyouts are typically in roles the company has already decided it can operate without. The mechanism is softer. The result is similar.
The timing of both announcements in the same week is not a coincidence in the broader pattern, even if each company made its decision independently.
What Is $700 Billion in AI Spending?
The Big 4, Alphabet, Microsoft, Meta, and Amazon, are projected to spend approximately $700 billion on AI-related capital expenditure in 2026. That's nearly double the previous year.
Where the money goes:
Data centers. New facilities optimized for AI workloads. These aren't traditional server farms. They're built around GPU clusters, liquid cooling, and power infrastructure designed for the energy demands of model training and inference.
Chips. NVIDIA GPUs (H100, H200, B200 series) plus custom silicon from each company (Google's TPUs, Amazon's Trainium, Meta's MTIA). The chip bill alone is in the hundreds of billions.
Training compute. The cost of training the next generation of models. Each model generation requires more compute than the last, and every Big 4 company is training its own frontier models.
Inference capacity. Running trained models at scale for billions of users. Every ChatGPT query, every Google AI Overview, every Meta AI response requires inference compute. As usage grows, so does the infrastructure needed to serve it.
$700 billion is not a research budget. It's an industrial buildout. These companies are constructing the physical infrastructure for an AI-driven economy.
Why Cut People and Spend More at the Same Time?
This is the question at the center of the story.
The short answer: the money isn't being redirected from salaries to AI. It's being spent in addition to salaries that were already there. The layoffs reduce operating costs. The capex spending builds new infrastructure. They come from different budget lines and serve different strategic goals.
The longer answer: AI changes what companies need. A team of 50 engineers maintaining internal tools becomes a team of 10 engineers managing AI agents that maintain the same tools. The 40 people who leave aren't replaced by cheaper labor. They're replaced by compute. The company's total spending goes up (because AI infrastructure is expensive), but its headcount goes down (because fewer humans are needed for the same output).
The market rewards this pattern. When Snap announced 1,000 layoffs on April 15 and disclosed that AI agents generate 65% of new code, the stock rose 7% the same day. Meta's stock reacted positively after its announcement as well. Wall Street is interpreting these moves as efficiency improvements, not distress signals.
That creates a feedback loop. CEOs see that the market rewards AI-driven headcount reductions. Boards see it. Activist investors see it. The incentive to replicate the playbook increases with every positive stock reaction.
How Big Is the 2026 Wave?
According to Layoffs.fyi, 92,000+ tech workers have been laid off in 2026 as of April 24. The year isn't even half over.
Since 2020, the cumulative total is approximately 900,000 tech layoffs globally.
In a single week in April 2026:
- Snap: 1,000 (April 15)
- Meta: 8,000 (April 23)
- Microsoft: up to 8,750 buyouts (April, ongoing)
That's potentially 17,750 jobs from three companies in roughly ten days. Each announcement referenced AI as a factor in the decision.
Which Roles Get Hit First?
Based on the specific language in these announcements and what AI is already doing inside these companies, the roles most directly affected fall into a recognizable pattern.
Junior engineering. When AI agents write 65% of new code (as at Snap), the need for entry-level engineers who write boilerplate, fix bugs, and build internal tools decreases. Senior engineers who direct AI and review its output remain. The junior pipeline narrows.
Customer support. AI agents at Snap handle over 1 million customer queries per month. Meta and Microsoft have similar deployments. First-line support roles are being absorbed by AI at scale.
Content moderation. Meta employed thousands of content moderators. AI classification systems now handle the bulk of content screening, with humans reviewing edge cases.
Internal tooling. Teams that built and maintained internal dashboards, admin panels, and workflow tools are being reduced as AI generates and maintains these systems.
Marketing operations. Campaign setup, ad creative testing, performance reporting. These are structured, repeatable tasks that AI handles well. The strategic roles (deciding what to market and why) remain. The operational roles (executing the campaign mechanics) shrink.
Two Ways to Read This
There are two legitimate interpretations, and the data supports parts of both.
The structural view. AI is permanently replacing categories of labor. The jobs being cut are not coming back when the economy improves. They're gone because AI does them now. The $700B in capex is building the infrastructure to accelerate this replacement. Tech is the first sector. Others follow. This is a one-way structural shift in the relationship between capital and labor.
The cyclical view. Tech companies over-hired during the 2020-2021 boom. They're now correcting to sustainable headcount levels. AI is the justification, not the cause. Companies always cut during efficiency cycles and blame whatever technology is trendy. In the 2000s it was offshoring. In the 2010s it was "automation." Now it's AI. The hiring will resume when growth picks up.
The Conversation published an analysis arguing that AI is being used as a convenient narrative for cuts that would have happened anyway due to over-hiring. CNBC ran a headline on April 24 calling it an "AI-driven labor crisis."
Both framings contain truth. Some of these jobs would have been cut regardless of AI. Some of these jobs are genuinely gone because AI does them. Figuring out which is which requires watching what happens when these companies start hiring again, and whether the roles they hire for look anything like the ones they eliminated.
What Does This Mean for You?
If you work in tech: Look at what your company is spending on AI infrastructure and what it's cutting in headcount. If your role involves tasks that AI agents are already handling elsewhere (code generation, support queries, content review), that's not a guarantee you'll be affected, but it's a signal worth tracking.
If you're investing: The companies spending the most on AI capex are also the ones cutting the most aggressively. The market is treating this as positive. Whether that continues depends on whether the AI infrastructure actually produces returns. $700B in spending needs to generate revenue. If it does, the stocks keep climbing. If it doesn't, the correction will be significant.
If you're thinking about policy: There's currently no regulatory framework that addresses AI-driven workforce reduction at this scale. Severance requirements, retraining programs, notification periods. These were all designed for an era when layoffs were temporary and cyclical. If the structural view is correct, the policy toolkit needs updating.
The $700B is being spent. The 20,000 jobs are being cut. The market is applauding. Whether that applause lasts depends on what the infrastructure produces and who gets left behind.
Follow @easyai.ai for more breakdowns like this.
---
Sources
---
Want more?
Browse our prompt packs, guides, and automation tools.