On 31 March 2026, at six o’clock in the morning local time, an estimated 20,000 to 30,000 Oracle employees across the United States, India, Canada, Mexico and Uruguay opened their inboxes to a single email from “Oracle Leadership” telling them their role had been eliminated and that the day was their last working day. There was no warning from HR or their managers. Access to company systems was cut within the hour.

This was not a company in distress. In its most recent earnings, Oracle reported a 95% year-on-year jump in net income to USD 6.13 billion and remaining performance obligations — contracted future revenue — of USD 523 billion, up 433%. The cuts are happening because Oracle has committed to roughly USD 156 billion in AI data centre infrastructure spending and its current cash flow cannot comfortably absorb the bill. Headcount is the most flexible cost on the balance sheet, so headcount is what gets cut.

For Acquism’s clients across AI, data science and engineering recruitment in Europe, the Oracle story is not really about Oracle. It is about a pattern that is now visible across the entire technology sector — and is starting to spread beyond it.

Oracle by the numbers

The financial logic is brutally simple. TD Cowen estimates the cuts will hit between 20,000 and 30,000 people, roughly 18% of Oracle’s 162,000-strong global workforce. Internal teams in Revenue and Health Sciences and in SaaS and Virtual Operations Services have reportedly seen reductions of at least 30%. Several former employees told TIME magazine they had spent the previous year documenting their workflows to train the AI systems that subsequently replaced them.

Oracle is not an outlier — it is the template

Across the four largest US tech companies — Google, Amazon, Microsoft and Meta — combined planned capital expenditure for 2026 stands at approximately USD 725 billion, up 77% on 2025’s already record-breaking USD 410 billion. That is more than the entire global oil and gas industry spends on exploration. Inside each of these companies, the same calculation is being run: human salaries are the only major cost flexible enough to be cut quickly enough to partially offset the capex.

Total tech sector layoffs surpassed 92,000 in the first four months of 2026 alone, on top of 245,000 in 2025. Nikkei Asia attributed 47.9% of Q1 2026 tech layoffs directly to AI and automation. A Resume.org survey of 1,000 US hiring managers found that 55% expect layoffs in 2026, with 44% naming AI as a top driver.

So who is genuinely next?

Forecasting which specific company will move next is a fool’s errand, but the structural signals are remarkably consistent. Three patterns flag a company as a likely candidate for the next wave.

Pattern 1: Aggressive AI capex against constrained cash flow. This is the Oracle template. Any company committing tens of billions to data centres or model training without the underlying cash to cover it is, mathematically, going to look at headcount. Watch second-tier cloud players, large enterprise SaaS vendors and mid-cap AI infrastructure plays in particular.

Pattern 2: A core product with high routine-work exposure. Companies whose revenue depends on activities AI now does well — first-line customer support, repetitive content generation, basic legal review, manual QA, simple bookkeeping — face structural pressure regardless of their cash position. Many will not announce mass layoffs; they will simply stop replacing leavers.

Pattern 3: Sector contagion beyond pure tech. Banking, insurance, professional services and large consultancies are now openly rebuilding around AI. JPMorgan, Goldman Sachs, the Big Four accountancies and the management consultancies have all signalled meaningful headcount adjustments tied to internal AI rollouts. Telecoms (BT, Verizon) and retailers with large back-office functions are following the same pattern.

Where the risk lands — and where it doesn’t

The labour-market data tells a more complicated story than the layoff headlines suggest. At the same time as those record cuts, the United States had roughly 275,000 open AI-related job postings, with companies reporting a 92% year-on-year increase in hiring for AI roles and a 56% wage premium on the highest-demand specialisms. Bloomberg’s analysis suggests roughly half of AI-attributed layoffs result in the same roles being rehired — often offshore or at lower salaries — meaning a meaningful share of what is presented as a labour-reduction story is in reality a labour-repricing story.

What this means for recruiters and HR

Acquism’s view from inside European AI and engineering hiring is that the recruiting playbook needs to change in three concrete ways.

• Stop hiring for tasks; start hiring for judgement. The roles that survived Oracle, Amazon and Microsoft’s cuts are those where the value sits in the human’s ability to direct, review and take accountability for AI output — not in the human performing the underlying task. Job specifications should reflect that.
• Build internal AI-displacement maps before competitors force the issue. The most sensible HR teams in 2026 are auditing every role against the question: ‘If we deployed best-in-class AI tooling tomorrow, what would this person actually do that the model could not?’ That answer should drive workforce planning, reskilling and redeployment — not a 06:00 email.
• Rebuild junior pipelines around AI-supervision work. Junior data analyst and developer roles in their pre-2024 form are largely gone. The functional replacement — junior staff supervising AI output, building eval pipelines, doing security review — is in acute shortage and represents one of the strongest career-entry pathways available right now.

What this means for candidates

If you have been laid off in the current wave, or you are watching the cuts in your sector and wondering when your name comes up, the data points to a small number of clear bets.

• Move up the value stack. Roles defined by execution of repeatable tasks are the highest-risk category in 2026. Roles defined by judgement, system design, security, or relationship management are the lowest-risk.
• Become fluent — visibly, demonstrably fluent — in modern AI tooling. The 56% wage premium on AI-adjacent roles is real, and it is being paid to people who can prove they are productive with these tools, not people who say they are open to learning them.
• Diversify geographically. Some of the strongest AI and engineering opportunities Acquism is seeing in 2026 are in Switzerland, Germany, the Netherlands, Ireland and the Nordics — where regulated industries, particularly financial services and pharma, are hiring AI talent aggressively while US Big Tech is cutting.
• Negotiate hard on severance and unvested equity. Many Oracle employees discovered the hard way that decades of service do not translate into protected severance. If you are joining a company with a history of restructurings, treat severance terms as a negotiable item from day one.

The honest assessment

It is tempting, when reading about 30,000 Oracle layoffs, to frame this as a temporary correction or an industry-specific story. The data does not support that reading. Across at least eight of the largest companies in technology, with combined 2026 capex exceeding USD 725 billion, the same playbook is being executed: invest heavily in AI, identify workflows the new tools can absorb, then communicate the resulting reductions with explicit AI attribution. Snap’s 16% cut, Meta’s planned 20% reduction over the year, and Oracle’s 18% wave are all visible iterations of the same template.

The realistic question for any technology professional in 2026 is not whether their employer will engage with this template, but when — and what they have done in the meantime to make sure they are on the side of the equation that benefits from it. Acquism’s experience placing senior AI, ML and engineering talent across Europe in the last twelve months suggests there has rarely been a starker bifurcation in the labour market: serious shortage at the top of the value stack, real pressure across the middle, and structural displacement at the base.

Oracle was not the first wave. It will not be the last. But it has made the underlying logic impossible to ignore — and that, perversely, may be the most useful thing about it.