Amazon Is Cutting Roles and Growing Its Workforce. Both Can Be True.

Empty office workstation overlooking a generic warehouse with parcels and workers.
AI-generated illustration of different kinds of work within a large retailer; not an Amazon facility or its employees.


Whenever I hear that AI is creating jobs or destroying them, my next question is: which jobs, and for whom? Amazon is a useful case because a very large workforce can keep growing while particular teams face painful cuts.

That is not a contradiction. It is a reminder that a company’s headcount is a total, while a career is something much more specific.

The numbers need dates and definitions

Amazon’s second-quarter 2026 results report the following global workforce figures:

Quarter endFull-time and part-time employees
December 20251,576,000
March 20261,575,000
June 20261,595,000

Amazon’s definition excludes contractors and temporary personnel. These are global totals across the company, not its corporate workforce alone.

Separately, Amazon’s January restructuring announcement said approximately 16,000 roles would be affected. Beth Galetti described reducing organisational layers and bureaucracy while continuing to hire in strategic areas.

Put those statements together and the picture becomes more useful: a company can cut thousands of roles and still have a higher total headcount at a later reporting date. Neither number tells us that the people leaving became the people being hired.

Amazon has also been explicit about AI

In his message to employees about generative AI, Andy Jassy said the company would need fewer people for some existing jobs and more for other types of work. He also expected efficiency gains from AI to reduce Amazon’s total corporate workforce over the following years.

That is a management expectation about the future. It is not a breakdown proving that every subsequent redundancy was caused by an AI system taking over a task.

My reading is that several things can be happening together: organisational restructuring, changes in investment priorities, automation and new hiring. It is too tidy to force every movement into one explanation.

A net total cannot describe a transition

Imagine a business eliminating a team in one city while opening specialist positions elsewhere. Even if the numbers balance perfectly, the qualifications, salaries, working hours and relocation requirements might not. A reassuring company-wide total does very little for someone who cannot make that move.

That is why I am cautious about saying AI simply “moves jobs around”. Movement has a cost, and sometimes a particular role really does disappear. Equally, a round of cuts does not establish that employment everywhere will fall.

The evidence I want is more detailed: which functions lost positions, which gained them, how many affected employees moved internally, and what happened to pay and working conditions. A headcount table cannot answer those questions on its own.

What a useful response looks like

For someone trying to plan a career, I think the practical starting point is the work inside the role. Which tasks depend on repetitive processing? Which require understanding a customer, making a judgement, checking a result or taking responsibility when something goes wrong?

Learning an AI tool can be worthwhile when it improves a real part of that work. Paying for a course because it promises immunity from redundancy is a much less convincing proposition. No training programme controls an employer’s restructuring decisions.

For employers, a serious transition should include paid learning time, realistic internal vacancies and clear explanations of what is changing. “Reskill” is an easy instruction to give and a difficult project to complete without support.

I explored that gap in my earlier post on AI employment forecasts and career changes. Amazon makes the same point tangible: the interesting question is not just how many jobs exist at the end, but how people get from their current job to the next one.

Comments

Popular posts from this blog

Obsidian’s New Search: Less Friction for Notes in Two Languages

OPPO Find X9 Ultra vs Find X10 Pro Max: Is the Older Ultra Still Worth Buying?

OPPO Find X10 Technical Preview: Cameras, Design and Missing Specs