The Operational Excellence Tools Series | #49: Amazon Made 30,000 People Train the AI That Replaced Them Before They Left.

Welcome to the unique weekend article for the Loyal Fan subscribers-only edition.

This is the #49 article of The Operational Excellence Tools Series.

Outlines and Key Takeaways

Part 1 – Official Announcement

Part 2 – Background and Meaning

Part 3 – Analysis Through the Lens of Operational Excellence

Part 4 – Lessons for Businesses

Part 5 – Conclusion

PART 1: OFFICIAL INFORMATION

On January 28, 2026, Beth Galetti — Senior Vice President of People Experience and Technology at Amazon — sent an internal memo announcing the layoff of 16,000 corporate employees globally, marking the largest workforce reduction in the company’s 30-year history. But this was not the beginning — it was phase two of a restructuring program that started in October 2025, when Amazon had already cut 14,000 corporate employees in the first round. In total, within three months from October 2025 to January 2026, Amazon eliminated approximately 30,000 corporate positions — equivalent to nearly 10% of the company’s entire corporate workforce. The hardest-hit departments included Amazon Web Services (cloud computing), Prime Video (entertainment), People Experience and Technology (HR), and core retail operations. Notably, the eliminated positions were not limited to administrative roles but extended to critical technical functions — employees supporting cloud services, analytics platforms, and customer delivery tools were all included.

CEO Andy Jassy had prepared the intellectual groundwork for these layoffs months in advance through a series of increasingly candid internal memos about AI’s role in replacing human labor. In June 2025, Jassy sent a company-wide memo containing the widely quoted statement: “As we roll out more Generative AI and agents, it should change the way our work is done. We will need fewer people doing some of the jobs that are being done today, and more people doing other types of jobs.” He added that “in the next few years, we expect that this will reduce our total corporate workforce.” By January 2026, when the 16,000 round was announced, Jassy escalated the language: “Many of these agents have yet to be built, but make no mistake, they’re coming, and coming fast.” He framed the restructuring not as cost-cutting but as cultural transformation — “reducing layers, increasing ownership, and removing bureaucracy” — with the vision of turning Amazon into “the world’s biggest startup.”

But what makes Amazon’s layoffs an unprecedented event in modern corporate history is not the 30,000 headcount — it is how Amazon required departing employees to “train” the AI replacing them before they left. According to multiple internal sources cited in media reports — although Amazon has neither confirmed nor denied publicly — employees slated for termination were required to participate in “knowledge transfer sessions,” during which they were asked to record and describe in detail their decision-making processes, creative workflows, and daily work methods. These recordings, along with prompt libraries generated during the transfer process, were then fed directly into training datasets to refine the AI agents that would assume the work they once performed. According to reports, many employees initially believed they were participating in a standard transition process — helping colleagues take over their work — without realizing that the “colleague taking over” was not a human but an algorithm. Entire engineering teams were reportedly replaced by automated workflows running on Anthropic’s Claude Sonnet platform, where small clusters of senior engineers use advanced AI models to manage workloads that previously required dozens of employees.

The clearest illustration of the “train the AI then leave” model is the Alexa division — once a flagship of Amazon’s hardware innovation. According to internal sources, the Alexa division was “hollowed out” from over 847 engineers to just 23 — a 97% reduction that, if confirmed, would rank among the largest single-division technical contractions in technology industry history. Hardware development was transferred to a team of 31 contractors in Bangalore, India, operating entirely with AI coding assistants to maintain operations at a fraction of the previous cost. This new model — which analysts call the “efficiency matrix” — prioritizes AI-augmented productivity over traditional headcount: within divisions like AWS, entire departments are reportedly being consolidated, with small groups of senior engineers using AI to manage workloads that once required dozens of people.

Regarding support for affected employees, Amazon provided a package including 90 days of full pay and benefits during the transition period, allowing employees to apply for other internal positions. Those who do not find or pursue new roles receive severance packages, continued health insurance, and outplacement services. However, it should be noted that reports of a third round — an additional 14,000 in Q2 2026 — remain officially unconfirmed. An Amazon spokesperson called these reports “false and not based in fact,” although multiple internal sources continue to confirm that documents circulating among senior management point toward further reductions. Regardless of the final number, what cannot be denied is that Amazon — with Jassy’s vision of “agentic workflows” and a “fewer people, more agents” workforce — is establishing a new operating model in which the line between knowledge transfer and self-elimination of one’s own position becomes nearly indistinguishable.