Tech firms cutting managers via AI, with staff wary of being test cases

Tech companies are dismantling layers of middle management, citing artificial intelligence as the catalyst that lets them achieve more with fewer supervisors. In recent months, a growing list of Silicon Valley giants — including Coinbase, Block, Amazon and Meta — have announced sweeping layoffs that explicitly target what they describe as unnecessary bureaucracy, replacing human chains of command with flatter, AI-driven structures.

The trend is unmistakable. At the end of 2025, openings for middle manager jobs in the United States had fallen by 42% compared with a peak in 2022, according to workforce data platform Revelio Labs. Given that managers accounted for 13% of the US workforce in 2022, the scale of the shift is substantial. While much of the public focus has been on US tech giants, the impact is global: some workers in the UK have begun exploring unionisation efforts over concerns about job security, workplace benefits and employee privacy as AI-driven changes accelerate.

Company after company flattens

Coinbase was the latest to act. Last week the cryptocurrency exchange laid off 14% of its workforce — around 700 employees — as part of a plan to become “AI-native”. Brian Armstrong, the chief executive, said in a tweet announcing the cuts that AI had “dramatically” changed the pace of work, leading to an “inflection point”. The company said it would eliminate “pure managers” and require all managers to directly contribute code and other work, with each manager expected to oversee 15 or more direct reports. “We’re fundamentally changing how we operate: rebuilding Coinbase as an intelligence, with humans around the edge aligning it,” Armstrong wrote.

Block, the fintech company formerly known as Square, went further. It laid off 40% of its workforce — more than 4,000 people — citing gains from AI. Internal organisation charts reviewed by the Guardian showed some engineering managers were assigned as many as 175 direct reports under the new structure. Jack Dorsey, Block’s chief executive, has said his long-term goal is to have all 6,000 employees reporting directly to him, eliminating management layers entirely. The company has split management duties: AI now handles information sharing between managers, reports and teams; “directly responsible individuals” oversee strategy and priorities; and “player-coaches” manage employee growth. “There is no need for a permanent middle management layer,” reads a joint statement from Dorsey and board member Roelof Botha.

Amazon has been trimming management layers since 2023. Two years ago, chief executive Andy Jassy told employees he planned to increase the ratio of employees to managers by at least 15% — a goal the company reached last year — to give workers a greater sense of ownership and reduce bureaucracy. Amazon has since cut 14,000 corporate jobs, with further layoffs announced in May 2026 affecting divisions such as Selling Partner Services. Jassy has suggested that AI means Amazon “will need fewer people” doing some jobs. Separately, data analysis suggests offshoring may also be a factor: Amazon has substantially increased its offshore job postings.

Meta announced a “year of efficiency” in 2023 under Mark Zuckerberg, who has said the company is “starting to see projects that used to require big teams now be accomplished by a single very talented person”. Meta is considering sweeping layoffs that could affect 20% or more of its workforce as it intensifies AI investment. Some employees have expressed concern that their own roles could be replaced by the AI systems they are helping to train.

How AI is reshaping the middle manager’s job

The common thread across these companies is a fundamental redefinition of what middle managers do. Anastassia Fedyk, assistant professor at the University of California, Berkeley’s Haas School of Business, has studied how AI is changing workforce composition. She notes that as AI tools make it possible to shift more work from managers to their reports, these structural changes could become permanent. “The middle manager role is about to be under a lot more pressure,” said Emily Rose McRae, an analyst at business and technology insights company Gartner who studies AI’s impact on the future of work.

Prateek Singh, a software development manager who left Meta at the end of April, described the pressure from inside the company. After he joined in June 2025, managers on certain teams saw their number of direct reports jump and were increasingly expected to contribute code — a departure from the traditional model where managers delegated and guided while individual contributors executed tasks. Singh said managers turned to AI tools to speed up drafting documents, consolidating notes and evaluating employees, and used AI to generate code. He switched his one-on-one meetings with his seven direct reports from weekly to every other week. In between, he communicated asynchronously using AI agents — bots that execute tasks without human intervention — that connected with his direct reports’ agents to collect updates and provide feedback.

“If managers are expected to either be writing a lot more code or have a lot more reports, what I see happening is more asynchronous, agent-driven management,” Singh said. He warned that this could erode the benefits of direct human interaction, such as mentorship, human judgment and guidance. At a competitive employer like Meta, where the battle to be a top performer “often feels like The Hunger Games”, he said AI cannot improve employee performance the same way humans can. He also saw a future in which managers under pressure might blindly submit flawed AI-generated suggestions, compounding errors across teams and leading to data leaks, security holes or system outages.

At Block, the shift has been even more dramatic. Freeland Abbott, a former technical lead at Square who was laid off in February, said managers previously operated with an estimated six to 12 direct reports. Under the new AI-oriented structure, some engineering managers have 175. Abbott worries that the human parts of the job — team motivation, connection and support — could slip through the cracks. “AI can’t provide team motivation, human connection or support in the way a person can,” he said. Offloading employee development to same-level colleagues could disadvantage less-experienced and marginalised teams, he added. Several former Block employees have responded with “Wow, thank God I was laid off”, Abbott admitted, though he acknowledged that laid-off employees are slightly “biased” against their former employer.

Risks, bottlenecks and the loss of human oversight

While the push to thin management ranks is gaining traction, experts caution that the benefits are far from certain. Raffaella Sadun, a Harvard professor who studies the future of work, said tech companies are “very well positioned to make these changes because they’re advanced from a tech perspective. But even so, they’ll have to incur the cost of change” — overhauling how work is coordinated, altering decision-making processes and shifting workers into different positions, including demotions. She noted that AI adoption requires strong leadership to communicate vision and create learning incentives, and that middle managers are crucial for translating strategy into action.

McRae pointed out that reducing the number of middle managers is likely to complicate an already stressful job. Gartner surveys indicate that many managers across industries would choose not to be managers again if given the choice. Fewer layers mean fewer opportunities for advancement, which could drive away human talent — potentially feeding tech companies’ goal of a leaner, AI-powered workforce.

Amalia Goodwin, global managing director at consulting firm Slalom, said simplifying management structures requires a complete redesign of how work gets done, giving more authority to lower levels to make bigger decisions alone. That, in turn, demands that companies provide employees with the resources, skills and training to judge between good and bad outcomes. As employee output increases with AI, Goodwin warned, work could slow down in unintended ways. If one team produces more with AI, the team that must approve all that work may be overwhelmed by volume. With fewer managers, it is critical to create structures that break down divides between units and keep information flowing.

Some experts remain deeply sceptical that the trend will last. Matthew Bidwell, management professor at the University of Pennsylvania’s Wharton School, noted a history of companies that have tried to break old hierarchies only to abandon the efforts or treat them as one-offs. Middle managers are in a “precarious” position because “it’s harder to define your value”, he said. But as tech companies experiment with fewer managers, they may find they are losing a level of necessary scrutiny. “It means one fewer layer of kicking the tires,” Bidwell said. “You’ll move faster, but you’ll break more things, and for some organisations that’s probably not the right trade-off.”

There is also evidence that AI may be used as a smokescreen for layoffs driven by other factors. Gartner data indicates that fewer than 1% of layoffs in 2025 were due to actual productivity gains from AI, suggesting a significant disconnect between the promise of AI and its proven impact. OpenAI chief executive Sam Altman and venture capitalist Marc Andreessen have separately suggested that AI might be used to mask reasons such as overstaffing or a slowing labour market. Meanwhile, Gartner predicts that by 2027, 75% of hiring processes will include AI proficiency testing, and that widespread anxiety about AI-driven job loss is undermining productivity and slowing adoption. “AI adoption is a culture issue, not just a training issue,” the firm noted.

For those inside the companies, the uncertainty is immediate. Singh, who felt his job at Meta could be at risk, chose to leave. Now working outside Silicon Valley, he said he is happy to watch from a distance. “It’s just too early in the experiment. I didn’t want to be the guinea pig.”