A self-sustaining model for the Cardano on-chain treasury

The treasury currently holds 1.621B ADA as inert balance. Under this component, it delegates that balance across a curated set of stake pools meeting published criteria: independent infrastructure, above-median pledge, public telemetry, mandatory auto-abstain registration, open-source contribution or operator tooling. DReps ratify the curated list at onset and review it annually.

At a conservative 4% net yield, the treasury generates approximately 65M ADA per year — roughly $15M USD at current prices — flowing back into the same treasury that's currently bleeding USD value. The Cardano Foundation's 2025 staking report shows the staked majority of ADA earning this yield while the on-chain treasury, the largest single ADA holder in existence, sits at zero.

Holding 100% of treasury in ADA isn't a neutral choice — it's a maximally-volatile portfolio that's lost 80% of its USD value in 18 months. This component allocates 10–15% of treasury to a structured reserve in BTC and USDCx, sized to cover approximately 18 months of operational costs in stable terms.

USDCx, live on Cardano since February 27, 2026, is Circle's first Tier 1 stablecoin natively issued on Cardano via xReserve infrastructure with full CCTP cross-chain interoperability. Tier 1 means USDC-backed, audited, and not experimental — a deliberate de-risking choice compared to algorithmic or synthetic alternatives.

When a project requests treasury capital, the default structure should be a repayable loan with milestone-based release and a return mechanism — not an irrevocable grant. Grants remain available where genuinely warranted (public-good infrastructure, research with no commercial path) but become the exception, not the rule.

This isn't a hypothetical: Snek Foundation's 5M ADA proposal, structured as a repayable loan, passed Constitutional Committee approval at 85.71%. The CC explicitly cited the loan structure as the reason for high approval. The pattern is already proven; we just need to make it default rather than exceptional.

The 2025 Net Change Limit of 350M ADA was 99.5% consumed. That's not discipline — that's a soft ceiling getting maxed out by default. This component proposes a harder annual cap at ≤220M ADA, established by Info Action and reviewable annually based on yield performance and ecosystem need.

The math is straightforward: at ~65M ADA/year in yield (Component 01) plus a 220M ADA spending cap, the treasury reaches sustained equilibrium. The depletion year disappears from the chart entirely. Without a cap, no amount of yield catches up to unrestricted spending.

An open-source dashboard at a stable URL publishes: epoch-by-epoch delegation distribution, pool performance, monthly yield returns to treasury, reserve composition, and full transaction history. All data is on-chain and independently verifiable. The dashboard code itself is MIT or Apache 2.0 — anyone can audit it, fork it, or run their own instance.

Real-time USD valuations are anchored by Pyth Network, live on Cardano since late 2025, with sub-100ms institutional-grade oracle feeds covering crypto, equities, FX, and commodities. This removes the most common objection to treasury reporting — that "USD value" is a moving target that someone has to define. Pyth defines it, transparently, with feeds from 125+ institutional publishers.

Under the current system, governance delegation is concentrated into a single active DRep choice per wallet. Delegators can re-delegate at any time, but the model still encourages concentration around individual representatives rather than distribution across multiple specialized voices.

A multi-DRep delegation model would allow holders to distribute voting power across multiple DReps based on domain expertise or policy preference, while preserving the same total voting weight.

Quarterly elections sound responsive in theory but become operationally heavy in practice — for voters, for DReps, for the infrastructure that has to support them. Annual elections hit the right cadence: rare enough that DReps can actually build a record between cycles, frequent enough that representation stays current with the ecosystem.

The critical design choice is what happens at each election: every delegation resets. This is not just "DReps re-stand for election" — every ADA holder's delegation expires and must be re-cast. There is no "set and forget" delegation that persists year to year unchanged. Every holder must re-evaluate, every year, whether their current DRep still represents them. Power becomes a function of current conviction, not historical inertia.

Voting data exists on-chain. DRep positions exist scattered across forums, X, blog posts. Voting alignment (did the DRep vote the way they said they would?) exists nowhere coherent. A public DRep performance ledger aggregates all of this: every vote cast, every position published, every alignment or deviation, every attendance — queryable, exportable, embeddable.

Building this site is the next concrete piece of work. It's open-source, audit-friendly, and explicitly community-governed (no single party controls the categorization of votes or the "judgment" of positions). The site itself becomes infrastructure, not a publication — anyone can query, anyone can build on top of it, anyone can fork it.

In traditional democracies, candidates campaign — they have to explain themselves before voters choose them. In Cardano DRep elections today, this is loose, optional, and most DReps win or lose with minimal public articulation of where they actually stand. The result: votes cast on name recognition, prior reputation, or random selection — not on positions.

Proposal: each election cycle has a mandatory two-week campaign window. DReps publish video position statements covering treasury policy, protocol direction, expected voting patterns on major governance categories, and conflicts of interest. The performance ledger (Reform 03) hosts and timestamps these statements, creating a permanent record voters can match against actual votes later.

Real political systems have removal mechanisms: impeachment, recall elections, votes of no confidence. Cardano DReps currently have none. A DRep who registers, attracts delegation, then never votes — or who actively disrupts governance — cannot be removed except by waiting out their term. This is a structural gap that lets bad actors collect power they never use, or use destructively.

The proposed mechanism is encoded in the Constitution via CIP, with explicit grounds (extended inactivity defined by missed votes, demonstrable abuse of position, conflict-of-interest violations), procedural safeguards (super-majority threshold, evidence requirements, public hearing period, right of response), and explicit prohibition against weaponization (no recall on policy disagreement alone, no anonymous initiation, no repeat attempts within a defined cooldown). The line is sharp: failing the job, yes; disagreeing with the majority, no.

The Connected Treasury Framework projects roughly 65M ADA per year in yield once operational. A portion of that yield — proposed at 5–10% — funds a DRep reward pool. Distribution is performance-based: voting participation, alignment between stated positions and actual votes, attendance and engagement in deliberation, and remaining in good standing (no successful recall actions) all factor in. A DRep who votes, attends, deliberates, and represents earns proportionally. A DRep who registers and disappears earns nothing.

This closes the loop. The CTF generates yield. The yield funds quality DRep participation. Quality DRep participation makes governance functional. Functional governance approves the kind of decisions — like the CTF itself — that grow the treasury further. The system pays itself by routing yield to the people doing the work of representing holders, not by extracting from those holders.