AI Layoffs in 2026: Why the Companies Spending the Most on AI Are Also Hiring the Most People

Recent AI layoffs are only part of the story. Companies investing heavily in AI have actually increased their workforce by about 10% in the two years following adoption, including entry-level positions. The reason for both the layoffs and subsequent rehiring is the same: AI shifts job value from pure task execution to judgment and ownership. While headlines suggest job security is dwindling, it's not as dire as it seems - you're likely safer if your role cannot be easily automated.

Notable examples include Block, Oracle and Meta, which cut thousands of jobs in 2026 while heavily investing in AI, but this is a supply-side view and not the full picture (The Guardian, Reuters). Working paper research combining spend data from Ramp with payroll records from Revelio Labs across over 21,000 US businesses found that companies spending heavily on AI grew their headcount by around 10% over two years (Ramp, June 30, 2026).

OpenAI's GDPval benchmark now tests models on 1,320 real-world work tasks across 44 occupations, each graded by professionals with 14+ years of experience (OpenAI, September 25, 2025). This benchmark demonstrates AI's capability to handle sophisticated, real-world tasks, which is the driving force behind company decisions to scale up.

The layoffs are concentrated in areas where AI investments are highest. For instance, Block laid off about 4,000 employees in February 2026 citing AI productivity gains, while Oracle reduced its workforce by 21,000 in 2026 fiscal year (13% drop) even as it invested heavily in AI data centers (Reuters, June 22, 2026). Meta also cut 8,000 roles around the same time as it shifted focus to AI infrastructure and hiring (TechRepublic).

Despite these cuts, the broader tech sector saw 121,516 employees laid off across 204 companies so far in 2026, a significant but still small percentage of total tech employment (Layoffs.fyi 2026 tracker, live figures as of August 2026). This is an important distinction to make.

AI has advanced to the point where it can now compete with experienced professionals on real tasks, not just exam questions. OpenAI's GDPval benchmark measures this by evaluating 1,320 specialized tasks spanning 44 occupations (OpenAI GDPval; arXiv:2510.04374, October 2025). Recent frontier models have achieved win-or-tie rates against human graders, indicating a major capability shift.

Additionally, a simulation by AI research lab Andon Labs showed that models from early 2025 often failed in a simulated vending machine business, while current leaders like Anthropic's Claude Opus 5 turned a $500 budget into thousands of dollars in simulated profit (Andon Labs, July 28, 2026). This represents a new level of multi-month, multi-decision economic competence for AI.

Companies are also finding that AI can dramatically increase productivity - if an AI tool doubles the output of a $100 employee, it makes sense for a company to invest more in AI rather than let go of their human workforce. This is why heavy AI adopters are actually hiring more people, as the data shows strong correlation between AI adoption and growth rather than shrinkage.

The paradox can be explained by three mechanisms. First, ownership cannot be automated - AI can issue mistakes or make poor judgments, but ultimately, a human must be held accountable. Companies scaling AI need more accountable humans, not fewer. Second, humans are needed to handle the exceptions that AI cannot address, such as novel, high-stakes, or emotionally charged situations.

In customer support, for example, AI handles repetitive, simple queries while humans tackle complex, emotional issues. The same pattern holds true in sales, where AI can qualify leads while humans focus on closing high-value deals. Lastly, the cheap intelligence provided by AI allows companies to pursue ambitious goals that were previously unaffordable, leading to increased hiring in certain areas.

So, while AI layoffs are real and concentrated in certain areas, the overall trend suggests a shift in job value and a need for more humans in roles that offer judgment, ownership and complex decision-making. Your job is less at risk than the headlines may lead you to believe, provided you operate in a role that cannot be easily automated.