Two Markets, One Job Title
Something fundamental has shifted in the AI talent market over the past six months. What used to be a single, undifferentiated pool of "AI talent" has split into two distinct markets with different supply dynamics, different compensation structures, and different hiring strategies.
Market 1: AI Researchers and Foundation Model Engineers. These are the people who understand model architectures, training dynamics, and the math behind generative AI. Demand for this talent remains intense but is concentrated among a small number of employers: model providers, major tech companies, and well-funded AI labs. This market is tight, expensive, and geographically concentrated.
Market 2: AI Application Engineers. These are the people who build products using models as components. They understand prompting, RAG, evaluation, orchestration, and production deployment. Demand for this talent is exploding across every industry, and the supply is growing rapidly as experienced software engineers retool.
The Bifurcation in Numbers
Market 1 compensation continues to escalate. Senior researchers at top AI labs are commanding $800K or more in total compensation. The pool is small and globally contested.
Market 2 compensation is stabilizing and, in some segments, declining. As more senior engineers develop AI application skills and as bootcamps and training programs produce graduates, supply is catching up to demand. Expect Market 2 compensation to look more like senior software engineering compensation than AI research compensation within 12 months.
- For most companies, Market 2 is the right market. Unless you are training foundation models, you need application engineers, not researchers. Paying researcher compensation for application engineering work is a misallocation of capital.
- Market 2 talent is more available than you think. Stop posting job descriptions that require ML PhDs for application roles. The best application engineers often come from strong software engineering backgrounds with 12 to 18 months of intense AI product building experience.
- Market 1 talent may not be effective in your environment. Researchers who thrive in lab settings often struggle in product environments where speed, user empathy, and cross-functional collaboration matter more than technical depth.
Hiring Market 1 talent for Market 2 problems is like hiring a Formula 1 engineer to fix your delivery truck. The skills are related but the context is completely different.
Adjust Your Strategy
Audit your open AI roles. For each one, determine whether it truly requires Market 1 capabilities or whether Market 2 capabilities are sufficient. Adjust job descriptions, screening criteria, and compensation accordingly. You will fill roles faster, spend less, and likely get better outcomes for your actual needs.