HLabs misses by a quarter of a point, then splits its request in two

0.23 points short

Pebble + Gerolamo - HLabs 2026 Budget asked the community for ₳8.0M for a year of work on three things: maintenance of the TypeScript tooling that, by its own account, most Cardano developers use directly or through libraries built on it, a node that runs inside a web browser, and a smart contract language. When its voting closed at the start of epoch 628 it stood at 66.77% of DRep power, 0.23 percentage points under the 67% a withdrawal needs, on 140 yes ballots, 43 no and 20 abstain. The shares in this edition are the yes power as a share of all counted power, with abstaining power left out and power that cast no ballot counted as no. It was the only thing decided in these three epochs.

The committee split six yes against one no. The Eastern Cardano Council, the one no, found that the request set aside no ada for periodic independent audits and oversight metrics, which the constitution requires of a withdrawal, with every other requirement met.

Why Yoroi and EMURGO abstained

Among the twelve largest DReps on its ballot list, ranked by delegated power at epoch 629, Yoroi W₳llet and EMURGO abstained, Sebastien Guillemot セバ and the Cardano Foundation voted no, and the other eight voted yes. The abstentions were about the node. Yoroi wrote that its conversations with developers had raised “genuine questions about whether TypeScript is the right foundation for a production-grade node”, that with “Amaru and Dingo already funded and progressing” a third alternative node was not yet the best use of treasury money in this cycle, that the ecosystem “does not require a new smart contract language at this point in time”, and that at over 8 million ada the scale of the request warranted reconsideration. EMURGO wrote the same on all three counts and added that it “would warmly welcome the team to resubmit with a revised cost”.

The Foundation voted no while backing Pebble and the maintenance of the TypeScript libraries, writing that “the combined proposal’s costs and uncertain practical value of the Gerolamo browser node prevent our approval”, and that it would “highly encourage a leaner withdrawal specifically for Pebble”. Sebastien Guillemot’s no was about the language: an alternative to Aiken “that’s 50% better” would not be the difference maker for adoption.

The yes votes were about demand. YUTA wrote that it had built a simple demo with both products and saw significant potential demand for them. The Eternl DRep Committee wrote that it had revised its vote on Dingo from no to yes and that, for node diversity, it would support the Haskell, Rust, Go and TypeScript implementations alike.

One epoch later, two requests instead of one

Pebble & Ecosystem maintenance: TypeScript core of Cardano, from the same firm, was submitted in epoch 629, one epoch after the first request expired, for ₳4.6M. It funds two tracks, the language and the maintenance of the TypeScript tooling, the two parts the Foundation had said it would support. A second request came in the same epoch, another ₳4.6M for the browser node on its own, and each of the two is voted on independently. Together they ask ₳9.2M where the expired request asked ₳8.0M, so the refiling is a split rather than a cut. The node the two abstainers had questioned now stands alone, and the language and tooling no longer depend on it.

The revised Pebble and tooling request values five full time equivalents at 1,150,000 dollars with a 15% contingency. The original 2,812,500 dollar request, ten full time equivalents with a 25% refundable contingency, also covered the browser node, which now has a request of its own, so the two dollar figures are not the same work at two prices. The exchange rate changed as well: thirty five cents per ada in the expired request, a method it says it borrowed from the Amaru node proposal, twenty five cents in the new one. It also adds a periodic independent audit of how the money is used, the item the Eastern Cardano Council’s no vote had named. Its path is in the review of epochs 633 to 635, and the browser node’s in the review of epochs 636 to 638.

Eight new requests for ₳97.2M

Eight new requests were submitted in epochs 627 to 629, asking for ₳97.2M between them, and all eight were still open when the window closed. Revised requests were returning to the ballot before the original had expired. Revised Cardano Summit 2026 Singapore came in epoch 627, ₳7.8M from the Cardano Foundation for that one conference on its own. Its text is precise about what changed: the request is decoupled from the sponsorship of a second conference, and the summit budget is cut by 22%, which it puts at 550,000 dollars, by going from three stages to two and by raising the Foundation’s own contribution from 250,000 to 380,000 dollars. The 22% is measured against the summit part of the earlier request, not against the ₳14.1M the bundle asked for in total. The conference request came back split too, like the HLabs one.

The bundle it was written to replace, Cardano Summit 2026 and TOKEN2049 Singapore, reached its own deadline exactly at the close of the window, at 10.0% of DRep power as of that close.

The largest of the eight was a ₳39.8M request from Tweag for two years of core infrastructure work, seventeen work packages with the Peras finality upgrade as its primary focus, submitted in epoch 628, and the second largest Cardano Vision 2026: Human Centred, Scalable, Post Quantum Secure - IO Research at ₳32.9M, submitted in epoch 629. That one buys a year of research rather than software: its document puts the sum at 7.9 million dollars for forty two outputs across seven work packages, papers, problem statements, prototypes and improvement proposals, from a consortium of nine research partners with thirty six full time equivalents, under three priorities it names as usability and governance participation, scalability, and post quantum security.

Nothing paid

The treasury paid nothing in the window. Between the readings at epoch 627 and epoch 629 its balance rose by ₳7.6M, from ₳1.62B to ₳1.63B. DReps cast 1,159 ballots: 374 in epoch 627, 299 in epoch 628 and 486 in epoch 629. Committee members cast another 17. Those totals include superseded votes, and 153 DReps cast a final ballot somewhere in the window. No stake pool ballot fell in it.