Cardano (ADA) sustainability report

Cardano is present on the following networks: Cardano.

Cardano runs on Ouroboros Praos, a proof-of-stake protocol whose security argument rests on a published line of peer-reviewed research rather than on computational work. Time is cut into epochs of five days, and each epoch into one-second slots. For every slot a private lottery decides whether a given stake pool is entitled to produce a block: the pool evaluates a verifiable random function against the epoch's randomness seed and its share of delegated stake, and a winning pool can prove its entitlement to everyone else without any prior announcement. Because the outcome is private until the block appears, an adversary cannot know in advance which operator to attack. Most slots pass empty, which is expected rather than a fault.

Stake is delegated, not locked. A holder registers a staking credential and points it at a pool of their choosing, and the funds stay in their own wallet, spendable at any moment, with no bonding period and nothing to wait out when they change their mind. Pools are registered on-chain with a declared margin, a fixed cost and a pledge of the operator's own holdings, all publicly readable. Selection of a slot leader is weighted by the stake sitting behind a pool at the epoch snapshot.

The protocol has no mechanism for confiscating stake. There is no slashing of either delegated funds or an operator's pledge: a pool that is offline, misconfigured or simply unlucky forfeits the rewards it would have earned, and that forgone reward is the entire penalty the system imposes. Settlement is probabilistic and strengthens with chain density as blocks accumulate, so the longer a block has been buried the more expensive it becomes to displace. Governance decisions now run on-chain through delegated representatives, an elected constitutional committee and pool operator voting, and a successor consensus design intended to lift throughput by separating transaction distribution from block confirmation is in development, with a fallback to the present behavior when the network is congested or under attack.

Cardano is present on the following networks: Cardano.

Block production is paid from two pools of value. A reserve releases a fraction of its remaining balance each epoch, and transaction fees collected during the epoch are added to it. The combined amount is shared among pools in proportion to the stake that produced blocks, with a portion diverted to the treasury that funds on-chain proposals. Within a pool, the operator's declared fixed cost is taken first, then the operator's percentage margin, and whatever remains is divided among delegators in proportion to the stake each contributed. Everything about that split is published on-chain before anyone delegates, so the terms are visible in advance.

Two design parameters shape where stake settles. A saturation threshold caps the rewards any single pool can earn once the stake behind it passes a share of the total, so oversized pools return less per unit of stake and delegators have a standing reason to move elsewhere. A pledge influence factor pays slightly more to pools whose operators have committed their own funds, which raises the cost of running many pools with nothing at stake. Neither mechanism takes anything away; both work by making one choice more rewarding than another.

Users face a transaction fee calculated from a published formula: a constant plus a per-byte charge on the serialized transaction, so cost follows size and is predictable before submission. Every output must carry a minimum quantity of the native asset in proportion to the size of the entry it creates, which prices the long-term burden a transaction places on the state that every node must hold. Script execution is metered separately in two dimensions, memory and computational steps, each with its own price, and the submitter sets a budget that is charged whether or not the script succeeds. Registering a staking credential and submitting a governance action both require deposits, refundable when the registration is retired or the action concludes.

Cardano is present on the following networks: Cardano.

The figure is built from the bottom of the network upward rather than inferred from the value of block rewards, because a stake-based protocol gives block production no incentive to consume more electricity as rewards rise. The starting point is the population of machines that actually run the chain. Pool registration is itself an on-chain record, so the number of registered producing pools is directly readable rather than guessed at, and each production node is customarily fronted by relay nodes that carry network traffic, which has to be accounted for as a multiplier on the count that registration alone reveals. Public node listings and network crawlers supply an estimate of that multiplier and of the additional full nodes run by exchanges, explorers and applications that hold a copy of the ledger without ever producing a block.

Representative hardware follows from the published requirements for running the node software, which state memory, storage and processor expectations. Measured power draw for machines of that description, taken under load and at idle, is applied across the estimated population, and idle draw matters disproportionately here: a node spends the overwhelming majority of slots doing nothing but validating and gossiping, so average consumption sits much closer to the idle figure than to the peak.

Several honest limitations come with that approach. The node population is observed from the outside and cannot be audited; operators running several nodes on shared hardware, or several chains on one machine, are counted in ways that no external observer can fully resolve. The hardware mix is a distribution approximated by a representative device. Where evidence is thin the assumptions chosen are the conservative ones, more likely to overstate the total than to understate it, and estimates are revised as observation improves. Throughput does not change the answer much, since a node's draw is dominated by being switched on rather than by the transactions it happens to process.

Cardano is present on the following networks: Cardano.

The renewable share is not measured at any node; it is inferred from where the machines are and what the electricity in those places is generated from. Location evidence for this network is better than for many, because stake pools advertise themselves to attract delegation: registration metadata, operator-published relay endpoints and public network crawls together give a usable picture of the countries and regions in which producing and relaying nodes sit. That picture is incomplete, as hosting providers obscure the physical location of a share of nodes and some operators deliberately publish nothing, so the observed distribution is treated as a sample and extended to the unobserved remainder rather than assumed to cover everything.

Each located node is then matched to published statistics for the grid that supplies it, and the renewable proportion reported here is the weighted average of those grid figures across the estimated node population, weighted by the consumption attributed to each location rather than by node count alone. Grid statistics are annual and regional, so short-term variation in how a particular facility is supplied is invisible to this method. Contractual arrangements such as renewable supply agreements are not counted, because the method describes the physical grid mix a node draws from, and a purchase contract does not change the electrons delivered at a given moment.

Energy intensity expresses the consumption attributable to one additional transaction. It is computed as total consumption over the period divided by the transactions settled in the same period. For a stake-based chain that quotient should be read with care in one specific respect: the denominator moves with demand while the numerator barely does, since a node draws close to the same power whether the chain is busy or idle. Intensity therefore falls as the network is used more, and the figure describes an average cost of throughput rather than the incremental energy a single extra transaction actually causes. Source data is processed by Our World in Data from Ember and the Energy Institute's Statistical Review of World Energy: Share of electricity generated by renewables.

Cardano is present on the following networks: Cardano.

Emissions are derived from the consumption estimate rather than observed directly, by attaching a carbon intensity to each unit of electricity according to where that electricity was drawn. The node locations established for the renewable calculation are reused, each is matched to a published carbon intensity for the relevant grid, and the product of consumption and intensity is summed across the estimated node population. No emissions factor is applied to the manufacture of hardware or to the construction of the facilities housing it; the boundary is operational electricity only, and embodied emissions from equipment production are outside it.

The two scopes behave very differently for a network of this shape. Scope 1 covers emissions from sources under the direct control of the operators — in practice, on-site combustion such as a backup generator running during an outage. Across a population of ordinary servers in commercial hosting facilities this is a negligible contributor and is reported as such rather than modeled in detail. Scope 2 covers the emissions embodied in purchased electricity, and for this network it is effectively the whole of the footprint, because the activity being measured is machines drawing power from public grids.

Greenhouse gas intensity per transaction is formed the same way as energy intensity, by dividing period emissions by transactions settled in the period, and carries the same caveat: consumption is close to fixed with respect to throughput, so the quotient describes an average rather than the marginal emissions of one more transfer. Uncertainty compounds across the chain of estimates, since an error in the node population propagates into consumption and from there into emissions, and grid intensity is itself an annual average that smooths over substantial daily and seasonal variation in how a grid is supplied. Figures are restated each period as node observation and grid data are updated. Carbon intensity data is processed by Our World in Data from Ember and the Energy Institute's Statistical Review of World Energy, and is made available under a Creative Commons BY 4.0 license: Carbon intensity of electricity generation.