Tokenomics records that 20% of protocol issuance is being redirected away from indexing rewards. It does not say what the money is for. This entry does, and it is a good exercise in a skill this ecosystem demands constantly: telling what an organisation has published from what it has merely said.
What changed
On 2026-08-19 the Foundation published A New Mandate for The Graph Foundation, updated on 2026-08-25. The change it describes is one of role rather than degree. Instead of “coordinating and underwriting third-party roadmaps from a distance”, the post argues the Foundation “must take direct responsibility for executing The Graph’s long-term vision”, which requires “organizational evolution, in-house technical capabilities, and dedicated resourcing”.
Read that as an org chart change with a budget attached, because that is what it is. The Foundation has historically held and deployed money while others built. It is now proposing to build.
The published priorities are five:
- Operational continuity, meaning in-house technical capacity for core infrastructure.
- Product acceleration, meaning Subgraph Studio becoming a unified platform.
- Enabling data service providers, meaning recruiting people to run them.
- Chain coverage and integration, taken into direct ownership.
- Vertical priorities: DeFi, institutional and enterprise, and AI and agentic applications.
What it costs, and who pays
The mandate post is unusually direct about the funding, and the sentence is worth having in front of you: “An active governance proposal from the Foundation seeks to redirect 20% of protocol issuance to support these expanded responsibilities.”
That proposal is GIP-0089, and it is the same 20% that reduces what reaches indexers, and through them delegators. So the mandate and the reward cut are one decision, not two, and anyone forming a view on either should be looking at both.
Two things to keep straight, because they are routinely merged.
Approval and activation are different events, and here they were six days apart. The Council approved GIP-0089 on 2026-08-26. The Issuance Allocator was not changed until Arbitrum One block 500701944 on 2026-09-01, which is when the Rewards Manager’s share became 96.584 GRT/block and the redirect began carrying value. The gap is closed, but the habit that caught it is the point: the tokenomics entry carries the calls, including the one contract that still reports the old figure.
A mandate is not a specification. GIP-0089 moves money. It does not bind anyone to deliver a particular thing with it, and there is no clause in it that returns the issuance if the work does not land. That is not an accusation, it is the shape of the instrument: the accountability here is reputational and governmental rather than mechanical.
The roadmap most people are discussing is not the one that was published
Here is the part worth slowing down for.
The programme is widely called Project Catalyst, and is usually described as eight numbered workstreams. Search the Foundation’s own publications for that name and you will not find it.
The eight items themselves, and what is actually blocking each one, are in Project Catalyst.
This is not a small distinction for anyone trying to follow the programme. Progress against a published roadmap is checkable by anyone. Progress against a list somebody transcribed from a call is checkable only by people who were on the call, and quietly becomes whatever the next call says it is.
Where independent assessments sit
There is at least one public attempt to score how much of the Catalyst roadmap existing community software already covers, published by The Night’s Watch on 2026-08-28 alongside an offer to collaborate.
Read it, and read it knowing what it is. It is an argument made by a party with an interest in the answer, about their own repositories, using a scoring method they chose and describe as editorial. That does not make it wrong, and the underlying repositories are public and MIT or Apache licensed so the claims are checkable. It does mean it is a position rather than a measurement, and this site is not the place that decides which. The same test applies to any coverage claim, including one from the Foundation.
The mandate change is also being argued about in the community on its merits, including whether the situation that produced it was avoidable. That discussion is on the forum and is not something this entry adjudicates.
Why the mandate changed at all
The mandate post describes what the Foundation will now do. It does not describe what prompted it, and the prompt is a Council vote: the decision not to designate Edge & Node as the Labs entity, after which Edge & Node stopped providing dedicated resourcing for protocol management and the Foundation stepped into that gap.
The Labs decision covers what that vote settled, what it did not, and the awkward fact that the fullest public explanation of it lives in a chat client rather than on a page.
Much of that argument turns on decisions whose reasoning was never published, which is a property of the process rather than of the parties. What the governance record does and does not show covers why, and what would have to change for it to be otherwise.
What to watch, concretely
- Whether the redirect is still flowing at the rate approved, which is a contract reading rather than an announcement. It activated on 2026-09-01. Tokenomics carries the calls, and the reason one of them still answers with the old number.
- Whether the eight items get published under a name and a URL. That is the single cheapest improvement available to this programme.
- Whether delivery is reported against them, on a cadence, in a place with a date on it.
- Who the in-house team turns out to be. Operational continuity is the priority everything else rests on, and it is a hiring problem rather than a protocol one. As of 2026-09-01 the Foundation put the count at five: a Network Infrastructure Lead, a Protocol Team Lead, two Senior Protocol Engineers and a Software Engineer, all described as long-standing contributors. That figure was given in Discord and appears on no Foundation page, which is the same weakness as the roadmap and is discussed in the Labs decision.
Before reading on: why does it matter whether the roadmap is a page rather than a call?
Because issuance is now attached to it.
While the Foundation was underwriting other people’s roadmaps, its own plans were its own business and a call was a perfectly reasonable place for them. GIP-0089 changes that relationship: a fifth of protocol issuance funds this work, which means every indexer and every delegator is paying for it whether they follow it or not.
Money from a governance vote comes with an expectation of legibility that money from a treasury does not. Not a contractual one, since the GIP contains no such clause, but a practical one: the people funding it should be able to read what it is, notice when it changes, and form a view without having attended anything.
The fix is cheap, which is the frustrating part. A page with the eight items, a date, and a status against each would cost an afternoon and would close the whole question.