Amazon’s HR chief last month sought to damp concern among employees that mass lay-offs were now the norm. Cuts of more than 30,000 workers since October were not the “beginning of a new rhythm”, insisted Beth Galetti. “That’s not our plan.”
The world’s largest company by revenue has enacted rounds of job cuts since chief executive Andy Jassy took charge in 2021 after a period of rapid expansion in the pandemic. The company’s leadership says the latest cull serves a strategic aim: to make Amazon operate “like the world’s largest start-up”, ready to innovate using AI and by running “with fewer layers and more ownership”.
Yet for survivors left to work another day, the mood has felt increasingly sombre. “Day to day it just feels untenable. Our workload is increasing and the number of [problems] to deal with is just piling up,” said one longstanding employee. “Some managers know this is the case, but executives just keep pointing to some bigger AI picture.”
As Amazon pushes to increase productivity and reduce headcount, pressure on the $2.2tn group’s workforce is growing. For competitors in tech and other sectors grappling with what may be the most consequential technological shift since the advent of the internet, the resulting tensions could be a warning of things to come.
Faced with aggressive competition from start-ups such as OpenAI and Anthropic, Amazon and other large tech companies are investing billions to develop AI and the data centre infrastructure it relies on. As executives weigh those investments with the realities of operating a profitable group, analysts expect job cuts and divestments will form part of a playbook for all large tech groups this year.
Jason Wong, an analyst at Gartner, said lay-offs were part of a strategy to free up capital to invest in technologies that promise potentially outsized returns. “Executives are asking: ‘How much money can we save to reallocate [into AI investments]?’”
But for leaner workforces set steeper targets, while employers plough cash into AI tools that might replace them, the result can be demoralisation and depletion. More than a dozen current and former Amazon employees spanning core ad, cloud and retail units said staff morale had slumped.
“There’s just been a shift to being lean and profitable,” said one senior AWS employee. “You’re being asked to achieve the same goals with a third of the people [in the team].”
Some workers said successive redundancies had affected morale and many experienced “survivor’s guilt”. Despite Jassy saying cuts would focus on “layers” of management that slowed down decision-making, some staff said Amazon had fired recent hires and “critically placed” senior managers.
Multiple Amazon engineers said their units now had to deal with a higher number of “Sev2s” — incidents requiring a rapid response to avoid product outages — each day as a result of cuts. Amazon disputed the claim that headcount cuts were responsible.
“There are people on my team who aren’t happy that we’re having to deal with this stuff,” said one Amazon engineer, adding the company was increasingly relying on sub-optimal solutions that create more problems. “From a product perspective, you’re just having to cut features and maintain technical debt,” they added, referring to increased costs or vulnerabilities that arise when technical fixes are integrated without proper planning.
Amazon has built and deployed several AI tools for staff, including its Kiro developer platform and Q chatbot. The company expects more than 80 per cent of developers to use AI tools at least once a week. It monitors employees’ AI adoption on an internal dashboard called Clarity, according to two people familiar with the matter.
However, multiple employees said such tools were not effective for complex tasks, although they had helped with ideation and early prototyping. “I’ve seen no evidence that AI is doing anyone’s job to the extent that we can lay someone off,” said one AWS engineer. “The only thing I’ve seen is people picking up the slack.”
Several employees pointed to one instance where they said the use of Kiro was partly responsible for a 13-hour service outage in December, after engineers allowed the AI tool to make changes and it opted to “delete and recreate the environment”.
“It’s unproven technology. Some experimentation is tolerable but the degree to which executives are enforcing its use is just out of step with reality,” said one senior AWS employee. Amazon said the involvement of AI in the December outage was a “coincidence”.
In a letter last year, Jassy told employees the group was benefiting from using generative AI, and that a further rollout of agents “should change the way our work is done”. Though it was hard to predict exactly how this would look, “in the next few years, we expect that this will reduce our total corporate workforce”, he wrote.
Amazon has told managers to incorporate projected AI efficiencies into headcount planning for the year, according to two people familiar with the matter.
Amazon’s headquarters in Seattle. Amazon is one of the few large tech groups to mandate a five-day return to office for corporate employees © David Ryder/Bloomberg
This has led to pushback, including an open letter signed anonymously by current and former staff members, which argued the company was investing in a future where it would be easier to discard employees.
“Here’s what we’re actually experiencing: higher expected output and shorter timelines, mandates to build AI tools for wasteful use cases, and massive investment in AI with little investment in career advancement,” the letter, published last November, said.
“As soon as AI got introduced, deadlines started shortening, and people were expected to do more as quickly,” said one signatory who was made redundant last year. “It was heavily implied we would be graded on use.”
Multiple current AWS software developers said they were being asked to take on new roles with the assistance of AI tools. They noted technical writing teams had been laid off, with engineers now required to complete these tasks. AI use had also been formally written into some workers’ promotion criteria, the developers said.
Amazon said it expected employees to “use resources — including AI — to make work more engaging, to build products that help our teams become more effective and to improve customers’ lives”. It said “understanding” how employees adopted new technology helped the company support them and share “operational efficiency gains” during review processes and throughout the year.
Anton Korinek, director of the University of Virginia’s Economics of Transformative AI initiative, said workers were justified in concerns that tools would create “new competitive pressures” and make some roles redundant. “In the medium term, I expect that AI will lead to very strong productivity gains in many white-collar professions, and that these will eventually be reflected in lower job numbers,” he said.
The threat of AI-driven job losses in a tighter labour market has already tipped the balance of power from staff to employers, with many workplaces hardening expectations. Amazon is one of the few large tech groups to mandate a five-day return to office for corporate employees and measures attendance with badge scanners.
Executives say this approach is to help staff work cohesively in a period of intensified competition. But some analysts are sceptical Amazon — an employer of more than 1.5mn, roughly a fifth in corporate roles — can operate as a lean start-up.
“A start-up is based on huge amounts of cohesion and commitment,” said Anna Tavis, a New York University professor and former HR executive. “There’s no way you can generate this level of performance if your actions [through lay-offs] say otherwise.”
Still, Amazon is committed to this approach. Galetti’s January memo to employees said the “world is changing quickly” due to AI and this would force the group to be more innovative.
She did not rule out further cuts. “We’re convinced that we need to be organised more leanly.”