The Solana Foundation is keeping its new validator delegation rules, and some validators will lose Foundation stake. Jacob Creech of the Solana Foundation set out the reasoning in a post on X on 10 November 2026, writing that the Solana Foundation Delegation Program (SFDP) "has done its job" and that the next stage of delegation will reward validators that improve network health.
The statement is the Foundation's first detailed answer since the rules were tightened on 6 November and since the independent validator collective Layer33 asked the Foundation to pause retirements. Creech did not announce a pause. "There will be validators removed from the original sfdp," he wrote in a reply an hour later. "We're making a bet we can do more for network health by discretionary staking."
Why Jacob Creech says the Solana Foundation Delegation Program has done its job
Creech described SFDP as built for three jobs: bootstrapping new validators, keeping a healthy testnet, and getting some validators to test changes early. "Testnet exists, validators often find issues early, and a large amount of validators have started from SFDP," he wrote. He listed the changes, which he said were "controversial", as:
- a 12-month limit on time in SFDP
- automatic graduation at 300,000 SOL
- a maximum of 100 validators
- a requirement to run your own validator
The Foundation's FAQ states the first two: membership lasts 12 months from a validator's first eligible mainnet epoch, and members retire once their stake from sources other than SFDP exceeds 300,000 SOL. The FAQ does not state a 100-validator maximum. It says 100 new validators are added to the stake bot each week. Creech's post lists the 100 maximum as one of the rules.
Creech said feedback had told the Foundation that SFDP "is no longer reactive enough nor provides enough criteria given the ever-changing landscape of Solana." SFDP will continue "for the newer validators," he wrote, while validators that "contribute greatly to network health should also be rewarded." He cited two precedents for delegation outside the program: Foundation stake to Blueshift, and to validators that were first to run Frankendancer, an early build of the Firedancer client. Blueshift's validator holds about 3.58 million SOL, around 0.83% of stake, according to Solana Compass validator data, though that data does not show how much of it came from the Foundation.
What happens to validators that already hold Foundation stake
Existing members can apply to stay. Creech wrote that smaller validators "are the lifeblood of the network" and "can submit their current contributions to be considered for continuation," adding in the same thread that "more than 100 have submitted already." The FAQ links a renewal form and a separate form asking validators to describe their contributions to the Solana ecosystem. Bloomingbit also reported the statement and the 100-plus submissions.
Asked whether the aim was to cut the number of validators, Creech replied: "No, but we know that may be a side effect if we cannot find enough good contributors." He said the rules are going into effect "slowly" and that "discussion is happening now."
The size of the program is where the accounts diverge. Creech wrote that "100 was more than the current remaining sfdp set, allowing us to grow." The Foundation's own dashboard showed 327 validators receiving Foundation stake at epoch 1080, about 47% of all validators, with 23.75 million SOL delegated, around 5% of staked SOL. svetlight74, an X user arguing for the small validators in the program, replied to Creech: "you currently have ~330 in the program. So my point that you're slashing ~200 small validators is true." That count of removals is svetlight74's claim, and Creech gave no number of his own. How the 100 figure relates to the 327 on the dashboard is one of the open questions below.
How Creech's response maps to Layer33's five asks
Layer33's letter speaks for independent validators, the group the retirement tranches affect, while the Foundation decides where its own stake goes. Here is how the Foundation's response lines up with what the Layer33 letter asked for:
| Layer33 ask | Foundation response on 10 November |
|---|---|
| Pause weekly retirement tranches until members are consulted | No pause announced; rules going in "slowly", validators "will be removed" |
| Six more months of delegation for Alpenglow community-cluster participants | Not addressed |
| Publish the analysis behind the 100 cap and 12-month figures | One line: 100 was more than the remaining SFDP set |
| Formal consultation before the rules are final | "Discussion is happening now" and a request for feedback; no formal process announced |
| A contributor model alongside the incubator model | Discretionary delegation for network-health contributors; existing members can submit contributions |
The last row is where the two sides are closest. Layer33 proposed rewarding uptime through upgrades, client and data-centre diversity and governance testing, and Creech described stake going to validators that "contribute greatly to network health."
Nakamoto coefficient 18 or 8: why the Solana Compass and Layer33 counts differ
Both figures measure how few parties it would take to control a critical share of stake. They count different parties against different thresholds. Network figures on our validators page put the Nakamoto coefficient at 18 at epoch 1082 on 9 November 2026: the fewest individual validators that together hold a third of active stake. Layer33's letter counts operators instead, grouping every validator a company runs, and measures against the more than 40% of stake needed to halt Solana under Alpenglow. On that basis Layer33 puts the figure at 8, and at 3 for the 20% safety threshold.
Grouping makes a large difference at the top. Figment runs two of the largest vote accounts in Solana Compass validator data, holding 3.89% and 2.13% of stake, about 6% together. A validator-level count treats them as two, and an operator-level count treats them as one. Neither count is wrong; they answer different questions. Layer33 used the operator-level figure to argue that stake is more concentrated than "the 18 or so that gets repeated in marketing" suggests.
Our figures do not separate Foundation-delegated stake, so they cannot show how SFDP retirements will move either count. So far the network looks steady: Solana Compass data counts 698 validators at epoch 1082, up from 694 at epoch 1078, with the top 10 holding about 24.6% of stake. Layer33's letter describes retirements running in weekly tranches, so a steady count a few days in says little either way.
SFDP open questions: criteria, timeline and removal count
Validators and stakers watching the changes will want answers on:
- how contributions will be scored, and who at the Foundation decides
- when the next retirement tranches run, and how many validators each removes
- how the 100-validator figure relates to the 327 validators on the Foundation dashboard
- where stake from retiring validators is redelegated
- whether Alpenglow community-cluster participants get extra time, as Layer33 asked
- whether the Foundation publishes the analysis or opens the formal consultation Layer33 requested
The validator count and concentration figures we track will show the first effects as tranches retire.
