Reading time: 3 min
Table of Contents
Key Takeaways
- Workforce swap: GM cut ~600 IT positions (over 10% of its IT department) and is actively hiring for AI-native roles, signaling a deliberate rebuild from the ground up — not just adding tools.
- What they’re hiring for: Agent development, data engineering, cloud-based engineering, prompt engineering, and model training — the gap between using AI and building with AI is the point.
- Structural shift: Layoffs followed executive departures and the hiring of a new CPO in May 2025; the automaker is prioritizing production-grade infrastructure over legacy IT.
The numbers behind the layoffs
General Motors cut about 600 salaried employees — more than 10% of its IT workforce. That’s not a downsizing story. It’s a skills swap. The company confirmed to TechCrunch that these are not all permanent headcount reductions. They’re making room for people who can actually build AI systems, not just add a AI tool to an existing process.
Let me be specific: GM is hiring for AI-native development, data engineering, cloud-based engineering, and agent and model development. That means prompt engineers, pipeline builders, and people who design agent workflows from scratch. This isn’t theory. The requirement is clear: they want people who build with AI from the ground up — not people who use AI as a side project.
Here’s what actually happens in production
Most enterprise AI adoption fails not because the models are bad, but because the teams aren’t structured to maintain the infrastructure. You hire a few machine learning engineers, bolt a chatbot onto a legacy system, and call it AI transformation. That’s not automation — that’s a liability. The demo worked. Production didn’t. Here’s why: you didn’t rebuild the workforce.
GM is doing the hard part. They’re clearing out workers whose skills no longer fit and deliberately recruiting for agent development, model engineering, and AI-native workflows. The real cost of keeping legacy talent who can’t work with modern orchestration stacks is time lost, incidents, and team dependency. That’s a failure mode most people get wrong.
The executive shakeup that made it possible
This didn’t happen in a vacuum. In May 2025, GM brought on Sterling Anderson as chief product officer — a veteran of Aurora and the autonomous vehicle space. He didn’t waste time. Three top software executives left in November: Baris Cetinok (senior VP of software), Dave Richardson (senior VP of engineering), and
Barak Turovsky (chief AI officer, who lasted just nine months).
GM filled those holes fast. Behrad Toghi (ex-Apple) joined in October as AI lead. Rashed Haq came on as VP of autonomous vehicles — he spent five years running AI and robotics at Cruise before GM shut it down. That’s a deliberate pattern: they’re pulling from companies that have actually put AI into production under real constraints.
August 2024 saw about 1,000 software workers cut. This is the second wave. The pattern is clear: GM is consolidation its technology businesses into one organization, and the people who don’t fit the new stack are being replaced with architects who understand agent orchestration, VPS deployment, and reliable automation pipelines — not demo-grade proof-of-concepts.
What this signals for the rest of enterprise
For the industry, GM’s restructuring is a reality check. Enterprise AI adoption isn’t about adding tools — it’s about rebuilding the workforce. The specific capabilities they’re hiring for show where demand is heading: agent development, model engineering, cloud-native architecture. That’s not a trend. That’s the new baseline.
Most people get this wrong. They think AI adoption is about training a team on existing tools. But the failure pattern is always structural — the people who built the old stack can’t maintain the new one. GM is acknowledging resource constraints and taking the incremental path: firing first, hiring smarter.
We built open source tools like OpenClaw and Hermes at Rebirth Distribution because the gap between demo and production is the only one that matters. This move by GM shows that even the largest enterprises are starting to think the same way: you can’t patch your way to production-grade AI. You have to rebuild.