PROTOCOL ECONOMICS — Solana × SIMD-0550

Solana's Inflation Now Decays Twice as Fast — and You Won't See It on the Day

7 min read
SolanaSIMD-0550InflationStaking

Solana's inflation schedule has been the same curve since 2021: start at 8% a year, shrink by 15% of itself every year, stop at 1.5%. Stakers have quoted "about 4-and-a-bit percent" for a while because that is roughly where the curve sits five years in.

SIMD-0550 changes one number in that curve — the annual shrink rate goes from 15% to 30% — and it does so in a way that guarantees nobody notices on activation day. The rate you see the day after is the rate you saw the day before. The entire effect is in the future, and the future arrives about twice as fast.

This post is the formula, the re-anchoring trick that hides the change, and a forward curve computed from the rate that is actually live today.


The Curve, as the Validator Computes It

The validator's inflation object has four parameters: an initial rate, a terminal rate, a taper, and a foundation share. The total rate in year t is:

rate(t)max( terminal, initial × (1 − taper)t )

With the defaults — initial 8%, terminal 1.5%, taper 0.15 — year zero is 8%, year one is 6.8%, year five is about 3.5%, and the floor of 1.5% is reached around year 10.5. The foundation took 5% of issuance for the first seven years; that period is over — getInflationRate reports a foundation share of 0 on mainnet today, so everything now goes to stakers and validators.

Worth knowing: "year" here is not a calendar year. It is slots since the inflation start slot divided by slots-per-year, and the start slot was aligned back one epoch when inflation was switched on. When slot times shrink, slots-per-year rises and the year counter keeps time correctly. The curve is defined in wall-clock terms even though it is evaluated in slots.


What SIMD-0550 Changes, and the Re-Anchor That Hides It

The feature sets the taper to 0.30. If that were all it did, the rate would drop sharply on activation, because 8% × 0.705 is a much smaller number than 8% × 0.855.

So the code does one more thing. At the epoch boundary where the feature activates, it reads the rate the old curve produces at that moment — call it the anchor — and rewrites initial so that the new curve passes through exactly the same point:

initial_newanchor ÷ (1 − 0.30)year

The rate is continuous at activation. There is no step. The only thing that changed is the slope going forward: from here on, each year is 70% of the last instead of 85%.

The implementation detail is a small piece of care: rather than mutating the shared inflation object, the node replaces it, so that every boundary bank anchors off the pre-activation schedule and sibling forks never see a half-applied change. It is the kind of thing that matters only if it goes wrong, and it is done right.


The Forward Curve, From Today's Live Rate

The table below anchors on the rate live on mainnet as this was published: 3.65%, read from getInflationRate at epoch 1035 on 15 September 2026. If you are reading this later, pull the current rate and rerun it — the shape is what matters, not this particular anchor.

Years after activationold taper 15% → new taper 30%
0 — activation day3.65% → 3.65% (no step)
13.10% → 2.55%
22.63% → 1.79%
32.24% → 1.50% (floor)
41.90% → 1.50%
51.62% → 1.50%
61.50% (floor) → 1.50%

From a 3.65% anchor the old curve takes about six years to reach the 1.5% floor; the new one takes about three. Summed across years one to three, the new curve pays about 27% less inflation than the old one would have — 5.84 percentage-points of issuance against 7.97.

Staking yield is not the inflation rate. It is the inflation rate scaled by the inverse of the staking ratio — with roughly 69% of supply staked, each staked SOL receives about 1.45× the headline rate — minus the validator's commission. The ratio and the commission are yours to look up; the halving of the slope is the protocol's.


Who Feels This, and When

  • Long-term stakers feel it as a slow leak. Nothing changes in the first epoch, or the first month. It shows up as a compounding gap against what the old schedule would have paid, and it is largest in years two and three.
  • Anyone modelling SOL yield against alternatives — liquid staking rates, lending rates, treasury bills — needs to rerun the crossover date. The date at which native staking stops out-earning an alternative moves earlier, potentially by years.
  • Validators feel it in revenue, which is why the same era brings block-revenue sharing and basis-point commissions: with inflation shrinking faster, the fee side of validator income gets relatively larger, and the tooling to split it with delegators is arriving at the same time.
  • Token holders who do not stake feel the opposite: dilution slows faster. Whether that is worth more than the lost yield depends on what you hold and why.

What This Post Cannot Tell You

It cannot tell you the activation date. The gate is one feature account, activated at an epoch boundary after 95% of stake adopts it, and the schedule is set by validator adoption rather than an announcement.

It cannot tell you your own yield, because that needs the staking ratio and your validator's commission on the day you ask.

And it cannot tell you whether 30% is the right taper. The proposal's argument is about long-run security budget and issuance; the code's job is to apply it without a jump. Both are doing exactly what they say.


Solana Inflation: Questions People Actually Ask

What does SIMD-0550 change?

The annual disinflation rate, from 15% to 30%. Each year’s inflation becomes 70% of the previous year’s instead of 85%, until the 1.5% floor.

Will the Solana inflation rate drop on activation day?

No. The code re-anchors the starting rate so the new curve passes through the rate in force at activation. The change is entirely in the slope from that point on.

How does SIMD-0550 affect Solana staking yield?

Yield tracks inflation scaled by the inverse of the staking ratio, minus commission. Over the following three years the new curve pays about 27% less inflation than the old one would have from the same anchor.

Is SIMD-0550 active on Solana mainnet?

It is one feature gate (double_disinflation_rate), activated at an epoch boundary after 95% stake adoption. Check a feature-gate tracker for the state per cluster.


The One Calculation to Do This Week

Take your staked SOL, the live inflation rate, your validator's commission, and the staking ratio. Compute your expected yield for each of the next three years under a 0.85 multiplier and under a 0.70 multiplier. The gap between the two columns is what SIMD-0550 costs you, and it is the only number in this post that is specific to you.

— The Gate Is One Account on Chain —

There is no announcement to wait for.

double_disinflation_rate activates at an epoch boundary once 95% of stake runs software that supports it. The upgrades tracker reads the feature-gate accounts from the chain every minute and shows the state on each cluster.

Open the Upgrades Tracker ↗

Mechanism read from the Agave validator source at commit beee69b958: feature-set/src/lib.rs and runtime/src/bank.rs. The anchor rate and the zero foundation share were read from mainnet getInflationRate at epoch 1035 on 15 September 2026; both columns of the table are arithmetic on that anchor.

xroot.devOn-chain research, published in full
𝕏 @xrootdev
← Back to all notes