March 25, 2026
Adapted from a Thelander panel with Jodie Thelander (J. Thelander Consulting) and Dan Grady of Fleming Martin, a boutique executive search firm placing leaders into top technology and life-sciences companies.
AI talent is moving faster than the market can keep up with — and compensation is straining to match. This panel paired real-time private-company pay data with an executive recruiter’s view from the negotiating table to unpack what AI engineers and data scientists are actually commanding, and how the terms around their equity are shifting.
Benchmark by Financing, and Watch for Global Competition
Total financing raised is the most reliable lens for compensation (series labels aren’t apples-to-apples — one company’s Series A is another’s Series C). Later-stage companies raising enormous rounds have reshaped dilution, key hires, and option pools, and companies are staying private longer with bigger initial rounds. The talent market is now global: the old US-only view is gone, replaced by worldwide competition for key roles, with the UK alone representing about half of the European data.
The AI Premium Is Real — and Large
AI engineers are hired at a significant premium. The prevailing theory is that AI companies need fewer engineers because productivity has jumped, but the lead talent commands a lot: even a first founding engineer (not a CTO) may get 2–5% equity. The cash premiums are striking — in the data, roles like data scientists carry a premium of $200K+ over standard market comp. One real example: a VP of AI search at a ~$200M-revenue company targeted $400K total cash, but the market was paying $800K to $1M; member-of-technical-staff and research scientists were still landing $500–600K. These premiums stack on top of already-strong base numbers.
Negotiation Levers Beyond Cash
Recruiters are seeing candidates negotiate well past salary. Key levers include a 10-year post-termination exercise window (versus the default 90 days, where a high exercise cost can otherwise wipe out your options), change-of-control protection (essential for executives), early-exercise rights on unvested options, and full severance protection. Severance typically runs 3–6 months, though aggressive negotiators and larger companies push CEOs to 12–18 months. A notable pattern: founder CEOs often don’t negotiate these terms for themselves — until they hire a number-two or number-three who does, prompting them to go back and match it.
A New Wrinkle: Token Grants
An emerging trend (covered in recent TechCrunch and FT pieces) is companies granting AI engineers and data scientists tokens — as much as $400,000 worth per year — to fuel their AI development. The message to the hire is: you’re valuable, you’ll burn a lot of tokens building, and we’ll make that investment in you.
Equity Terms Are Getting Aggressive — and Investors Are Pushing Back
RSUs are growing more popular as a predictable, consistent retention mechanism. For roughly 10–15% of jobs — mostly in data science — candidates with leverage are winning accelerated vesting with no cliffs: three-year schedules split 50/25/25 or 40/30/30, with the first year vesting quarterly. But a backlash is brewing: investors have been nearly unanimous that they dislike these aggressive terms. Through all of it, the employee option pool stays remarkably stable at 10–15%, with founders diluting and investors rising over time. Anti-dilution clauses get attempted but are rarely granted.
Merit Increases: Flat Outside AI
General merit increases have stayed flat at 2–4% across most roles over the past few years (CFO/finance and a few tech roles a bit higher). AI is the exception — it is “survival of the fittest.” Someone who genuinely knows AI, has proven they can boost productivity or save a company money, and has mastered the tools (prompt engineering, automation platforms, governance) commands a premium, because they’re the ones who will help the company actually use AI going forward.
The Advice: Learn AI, and Know What You Own
The recruiter’s closing advice to anyone in or around these roles: get to know AI — learn it, tinker, fail, build things. The power is remarkable (he described having a CTO friend build, in 15 minutes with three prompts, something that would have taken months). Get familiar with as many tools as possible — Claude, Perplexity, Replit, and others — because early adoption doesn’t guarantee a given tool survives, so breadth matters. And on the compensation side, the recurring principle holds: know not just what you own, but how you own it — because the AI market is pushing every lever, from cash to equity to change-of-control, severance, and vesting.
This article is adapted from a Thelander panel and is intended as general educational information, not legal, tax, or financial advice. For guidance specific to your situation, consult qualified compensation and legal advisors.