Equity refresh grants for engineers
Equity refresh grants for engineers
An engineer equity refresh is a follow-on equity grant on top of the original new-hire package, typically issued in years 2 to 4 to smooth the vesting cliff and stop senior-talent flight. In 2026, FAANG refreshes annually at 25 to 50 percent of original, Series B/C startups refresh in year 3 at 25 to 100 percent, and Series A founders mostly run ad-hoc. The refresh pool typically runs 3 to 5 percent of company dilution per year.
If you are a founder reading this and you have not modeled refresh dilution in your cap table, you are probably 18 months from losing your first senior engineer. This piece walks through the 2026 benchmarks, the math, and how to design a program that actually retains people.
Disclaimer: this is operational guidance, not legal or tax advice. Talk to a 409A appraiser and a startup attorney before you grant anything.
What an engineer equity refresh actually is
A refresh is a brand-new equity grant, on its own 4-year vesting clock, layered on top of an engineer's existing grant. It is not a re-pricing. It is not an extension of the original package. It is a separate award.
The whole thing exists because companies stay private longer now. The median time to IPO has stretched from 4 years in 1999 to roughly 12 years in 2026. That means an engineer who joined in year 1 with a standard 4-year vest finishes vesting in year 5, watches their unvested equity drop to zero, and then sits next to a new hire who just got a fresh 4-year package. Without a refresh, your year-5 engineer is effectively working for salary alone while everyone around them has the upside.
Refreshes are how mature private companies prevent that cliff. Carta data shows over half of employees at later-stage startups receive a refresh by year 2.
The 2026 refresh benchmarks: FAANG vs Series B/C vs Series A
The single most useful thing a founder can do this quarter is benchmark refresh size against the rest of the market. Here is what the 2026 data looks like.
| Stage | Refresh size (% of original) | Cadence | Pool dilution / year |
|---|---|---|---|
| FAANG (Google, Meta, Apple) | 25-50% | Annual | Already public, RSU funded |
| Series C+ ($100M+ raised) | 25-100% | Year 3 plus annual top-ups | 3-5% |
| Series B ($30-100M) | 25-75% | Year 3 trigger | 3-4% |
| Series A ($10-30M) | Under 20% | Ad-hoc | 1-2% |
| Pre-seed / seed | Generally none | Cash bonus instead | 0% |
A few things to notice. FAANG refreshes annually because the underlying RSUs are liquid; the engineer can sell on vest day. Series B and C founders have the harder math, because they are issuing options against an illiquid stock and still need the dilution to add up over a 4-year board horizon.
Series A is the messy middle. Most Series A founders refresh reactively, when an engineer mentions a competing offer, instead of building it into the comp framework. This is the worst of both worlds: you pay the dilution and you lose the goodwill, because the engineer knows they had to threaten to leave to get it.
The AI talent war has pushed all these numbers up roughly 15 to 25 percent year over year since 2024, especially for engineers with verifiable LLM-product shipping experience. If you are reading a comp breakdown for software engineer total compensation, refresh equity is the line item growing fastest.
The math: how a refresh program dilutes your cap table
Let us model a real example. You are a Series B with 50 employees, 30 of them engineers. Your board approved a 5 percent annual refresh pool. Year 3 hits and you grant refreshes.
- Total option pool consumed by refresh: 5 percent of fully-diluted
- Average refresh per engineer: 0.15 percent (5 percent across 30 engineers, weighted by performance)
- Founder dilution from this single refresh cycle: 5 percent
- Compound over 4 years if you keep the program annual: roughly 18 to 20 percent (less than 4x because of churn and FMV growth offsetting share count)
That is real dilution. It is also less than the cost of replacing those engineers. Industry consensus puts senior engineering replacement cost at 100 to 200 percent of annual salary, before counting the 3 to 6 month productivity ramp. For a senior at $200K base, that is $200K to $400K per loss, plus 3 months of stalled feature work.
So the trade is: 5 percent annual dilution against losing a quarter of your senior engineering team in any given year once vesting cliffs hit.
Evergreen vesting: the refresh pattern at later-stage startups
Stripe, Databricks, and most Anthropic-tier companies do not issue large biennial refreshes. They issue smaller annual grants such that every engineer always has 3 unvested years ahead of them. This is called evergreen vesting.
Mechanically: instead of granting 50 percent of original at year 3, you grant 12.5 percent of original every year starting year 1. After 4 years the engineer always has another 4 years of unvested equity rolling forward. The cliff never arrives.
For the engineer, this kills the year-3 anxiety. For the founder, the dilution is more predictable: 3 percent per year forever, instead of a 5 percent shock every other year. 88 percent of companies running refresh programs in 2026 tie the size to performance, so the actual grant varies by review tier (top performer, meets bar, below bar) but the cadence is fixed.
The downside: evergreen takes more board cycles. You need to defend a 3 percent annual pool every refresh. Most founders find that easier than defending a 10 percent biennial shock.
How to design a refresh program for your startup
Four decisions, in order:
1. Pool size. Most companies land at 3, 4, or 5 percent of fully-diluted per year, allocated to refresh. Lower if you are still pre-product-market-fit and need pool for new hires. Higher if you are cash-constrained and need equity to do the retention work cash would normally do.
2. Cadence. Three options: annual evergreen (best for retention, hardest on board cycles), year-3 trigger (most common at Series B), or performance-event-based (top tier only, every year). Pick one and write it into the comp doc.
3. Tiering. Top performer might get 1.5x base refresh, meets-bar gets 1x, below-bar gets 0.5x or 0. Make this transparent. Engineers compare notes. Any tier that feels arbitrary will create more attrition than it prevents.
4. Anchor. Anchor refresh size to dollar value at current 409A, not share count. As the share count expands and FMV grows, share-count-anchored grants quietly shrink in real terms. Pave's 2026 benchmarking covers 5,500 companies and almost all of them now anchor in dollars.
Calibrate against the broader question of how much equity to give a developer before you finalize the pool number, because new-hire and refresh pools compete for the same dilution budget.
When NOT to refresh (and what to do instead)
Refreshes are not free, and they are not always right.
- Pre-seed and seed. You probably cannot afford the dilution and your engineers cannot price the equity anyway. A cash bonus tied to a milestone (next round closing, first $1M ARR) typically retains better.
- 18 months from a likely exit. Refresh equity vests over 4 years. If you exit in 18 months, 75 percent of the refresh evaporates with no retention return. Acceleration clauses help but only partially.
- Project work, not strategic capability. If the work in question is a 12-week feature build, headcount equity is the wrong tool. Weekly contractor booking has zero dilution and zero severance risk.
This last point is where founders most often over-grant. You hire a senior engineer, give them a 0.4 percent grant, and 14 months later realize the work was a 6-month sprint, not a 4-year capability. The senior is bored, the equity is half-vested, and you cannot redeploy them without an equity discussion.
For genuinely strategic capability hires, refresh equity is correct. For project work, the math flips. A senior engineer on Cadence at $1,500 per week ships the same scope, costs $39,000 over 26 weeks, and ends cleanly with no cap table footprint. Every engineer on the platform is AI-native by default, vetted on Cursor, Claude Code, and Copilot fluency before they unlock bookings, so the velocity holds.
If you want to run that comparison against your current refresh budget, our ROI calculator takes a refresh dilution number and a contractor weekly rate and tells you the breakeven point.
Tooling: Carta, Pulley, and Pave for 2026
You will need software for this. The 2026 vendor lineup:
| Tool | Strength | Typical cost | Best for |
|---|---|---|---|
| Carta | Cap table + refresh modeling, board-pack templates, broad adoption | $3K-$15K/yr at scale | Series A through pre-IPO |
| Pulley | Faster board-approval flow, modern UI, cheaper | Roughly 30% under Carta | Sub-100 employee Series A/B |
| Pave | Comp benchmarking, refresh-size data across 5,500 companies | $10K-$30K/yr | Anyone designing the program from scratch |
Most founders run Carta or Pulley for the cap table and Pave for the benchmarking. The output of Pave (here is what a senior engineer at your stage gets in refresh) feeds the input of Carta or Pulley (here is the board approval pack to actually grant it).
If you are still negotiating the underlying term sheet as a non-technical founder, get the option pool sized correctly upfront. The refresh pool is part of that bigger pool conversation, and re-negotiating it later costs board capital you would rather spend on hiring.
What to do this quarter
Concrete sequence:
- Audit cliff timing. Run a list of every engineer 18 months or less from their original cliff. These are your refresh-or-lose population.
- Model the dilution. Pick 3 percent and 5 percent annual scenarios. Show the board what each does to founder ownership over 4 years. Carta and Pulley both ship templates for this.
- Decide cadence. Annual evergreen if you can defend it on the board. Year-3 trigger if not. Document the decision.
- Communicate proactively. The single biggest failure mode is doing refreshes reactively as a counter to a competing offer. By that point, the engineer already mentally left. Tell the team the program exists, the pool size, and the trigger.
The half-day of work to put this in writing is the highest-leverage retention move a Series B founder can make in 2026.
If you are sizing engineering spend against equity dilution and trying to decide between a refresh-locked headcount hire and a project-scoped contractor, run the numbers on Cadence. Senior engineers at $1,500 per week, weekly billing, replace any week, no cap table footprint. The 48-hour free trial means you only pay if the work clears the bar. It pairs well with an explicit equity negotiation framework for developer hires when you do decide to bring someone on full-time.
FAQ
When should we issue our first equity refresh grant?
Most startups issue the first refresh in year 2 or 3, anchored to the original cliff. If your engineer hits year 3 without a refresh and a competitor offers a fresh 4-year package, you have already lost the negotiation. Communicate the program before you need it as a counter-offer.
How much equity is in a typical refresh grant?
FAANG runs 25 to 50 percent of original annually. Series B/C runs 25 to 100 percent at year 3. Series A is usually under 20 percent and ad-hoc. The 2026 trend is smaller grants more often (evergreen vesting) rather than large biennial shocks.
What does a refresh program do to founder dilution?
A 3 to 5 percent annual refresh pool is standard at Series B and later. Over 4 years that compounds to roughly 12 to 18 percent in addition to new-hire dilution. Model it against the 100 to 200 percent of annual salary it costs to replace a senior engineer who left.
What is evergreen vesting?
Smaller annual refresh grants that always leave 3 unvested years ahead. Common at Stripe, Databricks, and Anthropic-tier companies. The engineer never feels the cliff, the founder gets predictable annual dilution instead of every-other-year shocks.
Should a Series A startup do refreshes?
Usually not on a fixed schedule. Most Series A founders refresh ad-hoc to retain a critical engineer who has grown into a senior role or to match a competing offer. Keep the pool below 20 percent of original until you have Series B clarity on dilution budget.