Is data science still a good career in 2026 or is the field saturated?

The entry level is crowded; the middle is starving. Both takes are true. What changed: 'I did a bootcamp and know pandas' no longer clears the bar — that layer is genuinely saturated, and LLMs automated chunks of it. What didn't change: companies are drowning in data and desperately short of people who can turn it into decisions. Where the demand moved: data engineering (pipelines, quality — chronically understaffed), ML engineering (models in production, now including LLM systems), and analysts who own a business domain deeply. If you're starting today: pick data engineering or ML engineering over generic 'data science', learn SQL to an expert level (still the most underrated skill in the field), and build projects on messy real data, not Titanic datasets. Saturation is a junior-generalist problem, not a field problem.