Cutting teams to pay for GPUs
What I'd tell an engineer worried about the next layoff round: what you can control, and what you can't.
When an engineer asks, “Could I be next?”, telling them to “just learn AI” feels inadequate.
TrueUp’s tracker shows more than 200,000 people affected by tech layoffs so far this year. Behind every number is someone having to rethink their plans.
This week, Canva announced restructuring affecting 22% of its workforce, including layoffs, role changes, and relocations. Reuters also reported that Google has been extending voluntary buyouts to more countries since June, amid heavy AI spending.
“Cutting teams to pay for GPUs” captures the anxiety. But the reasons behind individual restructures vary, and we shouldn’t turn every announcement into proof that AI has replaced those people.
What would I tell an engineer worried about being next?
First, being good at your job cannot guarantee that your job survives. Entire teams and priorities can disappear. A layoff is not a reliable verdict on someone’s ability.
Then I’d talk about where to build value.
In platform and UI modernization, writing the component is only part of the work. Someone has to understand the existing application contracts, access rules, dependencies, and workflows. Someone has to decide how to introduce change without breaking the applications other teams depend on.
An agent can help with implementation. The engineer still needs to establish what must remain true and demonstrate that the change works.
That is where I would focus:
- Understand the domain. Know the users, constraints, and consequences behind the requirements.
- Own work through to its outcome. Stay involved through integration, rollout, and the problems that surface afterwards.
- Learn to evaluate agent output. Practice spotting incorrect assumptions, missing cases, and changes that look convincing but fail in context.
- Make other engineers more effective. Share context, improve common components, and document decisions so the team can move with confidence.
- Explain your impact clearly. Connect your work to delivery, reliability, customer needs, or reduced maintenance effort.
I wouldn’t try to become indispensable by keeping knowledge to myself. I would aim to become someone the organization trusts with an unclear, consequential problem.
I’d also keep my CV current, maintain relationships outside my immediate team, and record concrete examples of my work. Preparation gives you options when organizational decisions are outside your control.
Working longer hours to prove your value is an exhausting response to uncertainty.
My advice would be to keep learning, get comfortable using agents, and build evidence that you can take a difficult problem through to a reliable outcome.
For those making the organizational decisions, the responsibility goes further: understand what expertise you’re removing, how the remaining team will develop, and who will own the consequences when something goes wrong.