The most talent-constrained field in the history of technology. A handful of frontier labs — OpenAI, Anthropic, Google DeepMind, Meta AI, Mistral — collectively employ a few thousand researchers who shape what every AI system on earth can do. The USA dominates at the frontier. The UK is the only other country with a genuine frontier lab presence, through Google DeepMind's London campus. France is building a serious challenger through Mistral. China runs a parallel AI ecosystem — DeepSeek, Baidu, SenseTime — separated from the Western stack by Nvidia export controls. For most graduates, AI careers means ML engineering at product companies, not research at frontier labs. That distinction matters enormously.
AI & Foundation Models splits into three tiers that look adjacent but are entirely different hiring markets.
OpenAI, Anthropic, Google DeepMind, Meta FAIR, Mistral. These labs hire from a narrow global pipeline: top-10 CS PhD programmes, NeurIPS/ICML/ICLR publication records, and direct research network referrals. A first-author paper at a top venue matters more than your degree institution. The number of frontier research roles that open globally each year is measured in the hundreds, not thousands.
the "AI engineer" or "ML engineer" role at cloud companies, enterprise software firms, and AI-adjacent startups. This is where the majority of CS graduates who want to work in AI actually end up. The hiring bar is high but not insane; the skills are more engineering than research.
every bank, retailer, logistics firm, and hospital now runs ML models. These roles are accessible, plentiful, and less prestigious. The most important planning decision in this sector is being honest about which tier you are aiming for and whether your academic profile realistically positions you for it. The USA is so dominant in tier-1 that studying anywhere else for a frontier research career requires explicit justification — UK (DeepMind) and France (Mistral, Meta FAIR Europe) are the only defensible alternatives.
See exactly which countries give you the best shot at these companies — scored, ranked, and sourced.