AI PM Equity Negotiation Guide: RSUs, Options, and Refresh Grants in 2026
TL;DR
Equity is where most AI PM offers leak value. FAANG hands you liquid RSUs with a 4-year vest and a refresh program; AI labs like OpenAI and Anthropic offer PPUs/profit interests that look enormous on paper but vest weirdly; early-stage startups offer ISOs that are usually worth zero. The negotiation playbook is different for each. This guide gives you the 2026 numbers, the scripts, and the math to know exactly when to push for equity, when to push for cash, and when to walk away from a vesting cliff that's structured to screw you.
FAANG vs Startup vs AI Lab: 2026 Equity Bands
Equity at an AI PM offer can range from $0 (small Series A startup) to $1.2M+ over four years (Anthropic L5). The number on the offer letter means almost nothing without context — vesting schedule, liquidity, refresh policy, and dilution risk all change what it's actually worth.
FAANG Senior PM (Google L5, Meta IC5, Amazon L6)
RSU grant: $400K–$700K over 4 years. Liquid stock, public market. Vesting: usually 25/25/25/25 or front-loaded (Meta does 25/25/25/25; Amazon does 5/15/40/40 — yes, that's a trap). Refresh: $80K–$180K/year after year 2.
OpenAI / Anthropic AI PM (L4–L5 equivalent)
OpenAI: PPUs (Profit Participation Units). Anthropic: profit interests. Grants: $400K–$1.2M over 4 years. Tender offers roughly annually for liquidity. Vesting often 4-year, 1-year cliff. Strike-equivalent dynamics differ — read the fine print on the cap.
Late-stage AI startup (Series C/D, $1B+ valuation)
ISOs or RSUs depending on stage. Grants: $250K–$600K paper value over 4 years. Liquidity: tender offers if you're lucky, otherwise wait for IPO/acquisition. Strike price matters massively at this stage — get the 409A and the latest preferred.
Early-stage AI startup (seed–Series B)
ISOs only. Grants: 0.1%–0.5% for senior PM, 0.05%–0.15% for mid-level. Strike price is low but so is exit probability. Realistic expected value: assume 80% chance the equity is worth $0. Negotiate cash first.
Big Tech AI division (MSFT AI, Google DeepMind, Amazon AGI)
Standard FAANG bands plus targeted AI premium of 10–25% for hard-to-fill roles. RSUs in the parent company stock. This is the lowest-risk equity in the AI PM market right now.
Vesting Cliffs and Schedules: Where the Trap Is
Two offers can have the same total dollar number and be wildly different jobs. The vesting schedule tells you when you actually get paid — and whether you'll feel locked in or free at month 18.
25/25/25/25 (Meta, Google, most startups)
Even quarterly vest after a 1-year cliff. You get 25% per year. Healthiest schedule. If a competitor pulls you in year 2, you walk away with 25–50% of the grant.
5/15/40/40 (Amazon)
Year 1: 5%. Year 2: 15%. Years 3 and 4: 40% each. Designed to keep you locked in. Amazon offsets early years with a sign-on bonus that drops off after year 2 — the year 3 income cliff is real. Negotiate the sign-on hard.
1-year cliff, then monthly
Standard at AI labs and many startups. You get nothing for 12 months, then 25% on the cliff date, then 1/48 per month. If you leave at month 11, you get $0. Plan accordingly.
No cliff, monthly from day 1
Rare and excellent. Some smaller companies offer this. Means even if the role doesn't work out at month 8, you've vested ~17% of the grant. Always ask whether the cliff is negotiable for senior hires.
Refresh Grants: The Compounding Layer Most PMs Underweight
The initial grant is what gets discussed at offer time. The refresh grant is what determines whether you stay rich or hit a cliff in year 4. Always ask about it in writing before signing.
What a refresh actually is
What it is: An additional equity grant given each year (or every 6 months at some companies) on top of your initial grant. Vests on its own 4-year schedule, layered over the original. The point: smooth out the year-4 income cliff that would otherwise hit when your initial grant fully vests.
Negotiation move: If a company doesn't have a refresh program, your year-5 comp is going to drop 40–60%. That's not a job you stay in — that's a job you re-negotiate every 4 years.
FAANG refresh expectations (2026)
What it is: Google: ~$80K–$150K/year refresh starting after year 1, performance-banded. Meta: $100K–$180K/year. Amazon: $60K–$120K/year, often tied to RSU performance bands. Apple: $80K–$140K/year, bonus pool. Microsoft: $80K–$160K/year.
Negotiation move: Get the refresh policy in writing. Recruiters will quote you ranges; you want "target refresh" language tied to performance ratings. Ask: "What was the average refresh for Meets-Plus performers at this level last year?"
AI lab refresh dynamics
What it is: OpenAI and Anthropic refresh annually based on tender valuations. Refresh size: 50–100% of initial grant value, depending on perf and tenure. Catch: refresh PPUs/profit interests come in at the new (higher) cap, so the absolute share count drops but dollar value grows.
Negotiation move: Don't just look at the absolute grant size. Look at refresh as a percentage of initial. A 15% refresh at OpenAI is normal; a 5% refresh is a soft layoff signal.
Negotiate Like You've Done It a Hundred Times
The AI PM Masterclass includes live negotiation coaching from a former Apple Group PM and Salesforce Sr. Director PM. Bring your offer; leave with a higher one.
When to Push for Cash vs Equity
Push for cash when the equity is illiquid and pre-IPO
Series A–C startups: discount the equity to 10–20% of paper value when comparing offers. If a startup offers $200K base + $300K paper equity vs FAANG $230K base + $150K liquid RSUs, the FAANG offer is better expected value. Demand cash to close the gap.
Push for equity when the company is at IPO escape velocity
Late-stage with confirmed IPO timeline (next 12–18 months): equity becomes effectively cash. Tilt the offer toward more RSUs, less base. Anthropic and OpenAI tender offers function similarly — you can convert paper to dollars roughly annually.
Push for sign-on when joining a 5/15/40/40 schedule
Amazon and Amazon-style schedules are designed to lock you in. Sign-on bonus across years 1 and 2 ($80K + $40K is typical for Sr. PM) compensates. If recruiter doesn't volunteer this, ask: "What's the standard 2-year sign-on offset?" It exists.
Push for double-trigger acceleration when joining late-stage
Means: if the company gets acquired AND you're terminated within 12 months, your unvested equity vests. Standard at most late-stage and required by sophisticated execs. Ask for it; many companies will agree if you push.
Push for early exercise on ISOs at early-stage startups
If strike is low, early-exercising your ISOs (with an 83(b) election filed within 30 days) starts the long-term capital gains clock. Tax treatment can be 20% LTCG vs 37% ordinary income on a successful exit. Worth thousands to millions on a real exit.
Real Scripts: What to Say When Negotiating Equity
Most PMs lose negotiations because they get vague. Specificity wins. Here are scripts that have closed real 2026 offers — adapt them to your situation, but don't soften them.
Asking for a higher initial RSU grant
"I'm enthusiastic about this role. Based on what I'm seeing in the market for AI PM Sr. roles at comparable companies — I have a competing offer at $X total comp — I'd need to see the equity portion increase by $120K over 4 years to make this work. Can we get there?"
Negotiating a sign-on to offset a back-loaded vest
"The 5/15/40/40 schedule means my year 1 and year 2 cash flow is meaningfully lower than the headline number. To bridge that, I'd want a $90K sign-on across years 1 and 2 — I've seen this offered for senior AI PM roles at this level."
Locking in refresh language
"Before I sign, I want to confirm the refresh program. What was the average refresh grant value for Meets-Plus rated PMs at this level last cycle? Can we get that target included in the offer letter?"
When the recruiter says 'equity isn't negotiable'
"I understand there are bands. I'm asking you to tell me where in the band this offer is, and what would move me to the top of the band. If we can't move on equity, I'll need a higher base or a sign-on to make this work."