The great displacement: How AI's workforce cull is fuelling the infrastructure boom

AI, he wrote in a memo made public via an SEC filing, was enabling smaller teams to “reduce repetitive work, increase velocity” and better support the company’s community, partners, and advertisers. By Wednesday afternoon, Snap’s stock had jumped roughly 8%.

Markets have stopped treating AI-driven job cuts as a warning sign. They have started treating them as a growth signal. And for the digital infrastructure industry, it is worth understanding exactly why.

The scale of Q1’s workforce reduction

Snap is not an outlier. According to data compiled by Nikkei Asia, 78,557 technology workers lost their jobs between January and April 2026: the highest quarterly total since early 2024, with more than 76% of affected positions in the United States. Of those cuts, 47.9% were attributed to reduced demand for human workers as a result of AI and workflow automation.

The largest individual actions tell the story. Amazon cut 16,000 corporate roles in January. Oracle is reported to have quietly cut around 30,000 positions, with savings said to be redirected toward data centre funding. Atlassian reduced headcount by 1,600 and simultaneously restructured its technology leadership around AI. Meta trimmed a further 1,500 from its Reality Labs division.

What distinguishes this wave from the post-pandemic corrections of 2023 and 2024 is the stated rationale.

The paradox at the heart of the story

Every role eliminated in the name of AI efficiency requires more of something else: compute, power, cooling, connectivity, and the physical buildings to house it all.

The hyperscalers leading the workforce reduction are the same organisations committing capital at a pace the industry has not seen before. Amazon has committed $200 billion in capital expenditure for 2026, up from $131.8 billion in 2025 with the majority directed at AWS and AI infrastructure, according to CEO Andy Jassy’s Q4 2025 earnings guidance.

Google’s guidance sits at $175 to $185 billion, up from $91 billion. Meta has outlined $115 to $135 billion. Microsoft is tracking toward $120 billion or more. Oracle has targeted $50 billion. Combined, according to Futurum Group, these five companies plan to spend between $660 billion and $690 billion on infrastructure in 2026, the vast majority directed at AI compute, data centres, and networking.

BloombergNEF places the total capital expenditure of the 14 largest publicly owned data centre operators globally at close to $750 billion for 2026, against roughly $450 billion last year. More than 23 gigawatts of data centre capacity was under construction globally at the end of September 2025. Between August 2025 and February 2026, analyst expectations for 2027 spending by those same operators were revised upward by 56%.

Nvidia chief executive Jensen Huang has estimated that between $3 trillion and $4 trillion will be spent on AI infrastructure by the end of the decade.

All of the major hyperscalers have reported that their markets are supply-constrained rather than demand-constrained meaning the limiting factor on AI infrastructure growth is not appetite, but the ability to build fast enough.

The power equation

The energy dimension is where pressure on the wider infrastructure ecosystem will be felt most acutely. According to a report from PowerLines, US investor-owned utility companies are planning to spend $1.4 trillion on electricity infrastructure by 2030, more than double the investment of the prior decade, driven in large part by data centre demand. That figure represents a 20% increase from the same utilities’ 2025 projections.

Deloitte’s 2026 Power and Utilities outlook estimates that data centre demand alone could reach 176 gigawatts by 2035, a fivefold increase from 2024 levels.

The hardware driving that demand is growing more power-hungry, not less. According to IoT Analytics’ Data Center Equipment and Infrastructure Market Report, modern AI server racks are already exceeding 100 kilowatts per rack — far beyond the 5 to 15 kilowatts that traditional CPU racks required. Peak rack density is projected to exceed 1,000 kilowatts by 2029.

A note of caution on attribution

Not every cut being labelled an AI layoff is straightforwardly that. Cognizant chief AI officer Babak Hodjat told Nikkei Asia that it would take another six months to a year “before companies start seeing real productivity gains from AI,” questioning whether current cuts are directly tied to actual productivity improvements.

OpenAI chief executive Sam Altman acknowledged at the India AI Impact Summit that “there’s some AI washing where people are blaming AI for layoffs that they would otherwise do.”

The true picture is probably more complex than the efficiency narrative suggests. But the infrastructure investment is real regardless of the motivation behind individual headcount decisions.

What it means for this industry

The technology sector’s workforce reduction and the infrastructure sector’s investment boom are, in financial terms, connected. Efficiency savings from AI-augmented operations are being recycled directly into the capital expenditure cycles driving demand for data centre capacity, power infrastructure, and connectivity.

For operators, developers, and carriers building and connecting that infrastructure, the demand signal has rarely been clearer. The constraints are not on the demand side. They are on the supply side: power availability, construction capacity, and the speed at which new data centre projects can be permitted and built.

Tech layoffs continue into 2025 as industry cuts another 26,000 jobs

OpenAI to double workforce to 8,000 amid industry layoffs