The Talent War Has Reached Interns: In AI, “Adult Money” Is Now Paid Right at the Entry Level
By the end of 2025, it became fully clear that the “talent wars” in AI are no longer just stories about $100 million bonuses for superstars. According to Business Insider, the largest players have begun paying students, PhD candidates, and junior researchers at levels that other industries reserve for experienced professionals. Internships in AI are no longer merely “entry-level positions” — they are turning into a strategic recruitment tool.
The most telling evidence is the numbers. Anthropic launched a four-month AI Safety Fellows program paying $3,850 per week, with a compute budget of up to $15,000 per month. OpenAI offers a six-month residency paying around $18,300 per month, where residents work as full-fledged contributors on research tasks. At Google DeepMind / Google Research, PhD students are hired as Student Researchers with annual compensation in the $113,000–$150,000 range, while Meta pays research interns $7,650–$12,000+ per month.
| Program / Company | Format | Compensation level (as reported) |
| Anthropic — AI Safety Fellows | 4 months | $3,850/week + up to $15,000/month for compute |
| OpenAI — Residency | 6 months | ~ $18,300/month |
| Google DeepMind / Google Research — Student Researcher | Annual | $113,000–$150,000/year |
| Meta — Research Interns | Monthly | $7,650–$12,000+/month |
Against this backdrop, the very top of the market looks just as revealing. Recruiters quoted by Business Insider mention base salaries of $300,000–$400,000 per year for leading AI specialists even at startups. And in U.S. H-1B filings linked to Mira Murati’s Thinking Machines Lab, base salaries of $450,000–$500,000 per year appeared for certain technical roles — before stock options and bonuses.
The point of this publication is not “everyone should rush into AI.” The point is that 2025 cemented a new norm: early talent has become a scarce resource, and companies are paying for it upfront — with money, compute, and near-employee status. This will both accelerate progress and deepen inequality: the best candidates will enter elite ecosystems earlier, while others will have to compete not only on skills, but also on access to infrastructure and mentorship.
