What the evidence actually says
Demand for data engineers has not declined since the widespread adoption of AI coding tools — it has grown. Companies are building more data infrastructure, not less, because AI features require large amounts of clean, accessible data to function. The demand for engineers who can build, maintain, and ensure quality of that data infrastructure is increasing alongside investment in AI.
The change is in how the work gets done. A data engineer using GitHub Copilot, ChatGPT for debugging, or AI-powered SQL optimisers can accomplish in hours what previously took days. That productivity multiplier does not eliminate the role — it means individual engineers can own more of the system and deliver more value.
The right way to think about this
The engineers most at risk are those doing purely repetitive, low-judgment work without developing understanding of the underlying systems. The engineers who are most valuable are those who understand distributed systems deeply enough to know when AI suggestions are wrong, who can design data architectures that will hold up as requirements change, and who can debug real production failures that no AI tool has seen before.
Learning data engineering in 2026 means learning it alongside AI tools, not in opposition to them. The goal is deep enough understanding that you can use AI to accelerate your work without relying on it to compensate for gaps in your knowledge.