Canton (CC) sustainability report

Canton is present on the following networks: Canton.

Canton is not a single replicated ledger and describing it as one would misstate how it works. It is a network of interoperable ledgers, each held by the participants entitled to see it, synchronized so that a transaction spanning several of them commits atomically or not at all. A given node stores only the data for transactions its operator is a party to, so there is no global copy of all activity held by every participant, and confidentiality is a structural property rather than something added on top.

Its consensus is described as proof of stakeholder, which differs fundamentally from proof of stake and the distinction matters. Validators are not selected by how much of an asset they hold. For each transaction, the parties to it are the ones who validate it and are responsible for confirming its effects, so the validating set is determined by who is involved rather than by a stake-weighted lottery. Confirmation runs as a two-phase commit: the relevant participants validate and confirm, while a synchronizer orders the requests and resolves conflicts between them.

That synchronizer is where a Byzantine fault tolerant agreement does operate, run by a distinguished class of node operators at a two-thirds threshold. Membership of that class is by invitation from the network's governing foundation rather than permissionless, and each such operator holds one vote in governance, with voting power equal among them rather than proportional to holdings. The published operator list names substantial financial market infrastructure and institutional participants and has grown through 2026 to several dozen entries.

The consequences for the security model should be stated directly. There is no anonymous validator population and no open entry to the role that orders transactions, so the network's integrity rests on the governance of that invited operator set rather than on the economic cost of attacking it. Published sources describing the network's reward mechanics disagree on whether a delegation or staking mechanism exists for ordinary holders, and this text makes no claim on that point.

Canton is present on the following networks: Canton.

Rewards are issued in rounds and directed at activity rather than at bonded capital. The operators of the synchronizing nodes earn for performing that function, and participants building applications on the network earn for the activity their applications generate, so the issuance flows toward the parties who operate infrastructure and who bring usage. A change taking effect during 2026 redirected issuance away from a category of reward previously available to ordinary validator operators and toward application and synchronizer operator rewards, which shifted the economics of running a node that merely stays available without transacting.

Participants obtain the capacity to transact by acquiring traffic on the synchronizer rather than by bidding in an open fee auction. This is a meaningful difference from a public chain's fee market: cost attaches to the throughput a participant reserves and consumes, which suits institutions that need predictable operating costs and to know in advance what participation will cost them, and it does not produce the congestion-driven price spikes that a public auction generates when demand rises sharply.

The penalty structure is where this network departs most sharply from proof-of-stake designs, and it should be described with care. The protocol does not secure itself by confiscating bonded capital, because participation does not require bonding capital in the first place. Node operators are identified institutions admitted by invitation, and the response to an operator that misbehaves or fails to perform runs through network governance and the agreements under which it was admitted, rather than through an automatic on-chain penalty. Whether any mechanism exists for ordinary holders to delegate an asset to an operator and share in rewards is described inconsistently across published sources, and no assertion is made here either way. What can be stated is that the network's integrity rests on the accountability of a known, invited operator set, and that accountability is institutional and contractual rather than cryptoeconomic.

Canton is present on the following networks: Canton.

This network inverts the hardest problem in estimating a blockchain's energy use, and the method changes accordingly. Elsewhere the dominant uncertainty is how many machines exist, because participants are anonymous and must be counted by crawling. Here the operators of the nodes that order transactions are admitted by invitation and published by name, and the list is maintained openly, so the population is enumerable rather than estimated. The count is a fact that can be read, not a figure inferred from network observation.

That shifts the uncertainty entirely onto the second question: what each named operator runs. These are financial institutions and market infrastructure providers operating to their own internal standards, typically with redundancy, failover and separate environments that an external observer cannot see, so the machines behind one named operator are likely to number more than one and the multiplier is not disclosed. The estimate applies a representative configuration derived from the published requirements for running the node software, together with an assumption about redundancy appropriate to operators of this kind, which is a stronger assumption than the hardware inference itself.

A second population must be added: the participant nodes operated by organizations that transact on the network without being part of the ordering set. Each holds only the data it is party to, so a participant node's storage burden is a fraction of what a full node on a globally replicated chain carries, and its resource requirements scale with that organization's own activity rather than with the network's total. The count of such nodes is less openly published than the ordering set and is estimated from disclosure and public listings.

The result is an estimate whose limitations differ in kind from those of a public chain. Population risk is low and hardware risk is high. Where evidence on redundancy and configuration is thin, conservative assumptions are used that are more likely to overstate than understate the total, and figures are revised as operators disclose more about how they run.

Canton is present on the following networks: Canton.

The geographic question is more tractable here than for any anonymous network. The operators of the ordering nodes are named institutions, and where a named financial institution runs infrastructure is generally either disclosed or inferable from its regulatory and operational footprint, so the distribution across jurisdictions rests on identification rather than on statistical inference from network observation. That is a materially stronger basis than for a network of pseudonymous validators, and it means the renewable share for this network can be stated with less geographic uncertainty than for most.

What remains uncertain is narrower but real. A named institution may run its nodes in commercial facilities in a jurisdiction other than its own headquarters, and while the operator is known the facility often is not. Institutions of this kind also commonly run redundant sites in separate locations for resilience, so a single operator may contribute consumption in more than one grid region in proportions that are not published. Participant nodes belonging to transacting organizations are less openly documented than the ordering set and their distribution is estimated with weaker methods.

Each location is matched to published statistics for the grid supplying it, and the renewable share reported is the average across those grids weighted by the consumption attributed to each location. A caveat specific to this network deserves stating: institutional operators frequently hold renewable supply agreements as part of their own environmental commitments, and those contractual arrangements are not counted here, because the method describes the physical grid mix from which a facility draws rather than a procurement position. This network's reported share may therefore sit below what its operators report under their own corporate accounting, and the difference is one of method rather than of fact.

Energy intensity per transaction is period consumption divided by transactions settled. The node population is fixed by invitation rather than responsive to usage, so consumption is close to fixed with respect to throughput and intensity falls as the network is used more. It is an average, not a marginal cost. 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.

Canton is present on the following networks: Canton.

Emissions are derived by applying a grid carbon intensity to the electricity attributed to each location in the distribution described above, across both the named ordering operators and the participant nodes of transacting organizations, and summing. The boundary covers operational electricity; manufacture of hardware and construction of facilities are outside it.

Scope 1 covers emissions from sources under the direct control of node operators. For institutional operators running in commercial or their own data facilities this means on-site combustion, principally backup generation during outages, and it remains a negligible contributor to the total. It is reported as such rather than modeled in detail, with the qualification that operators running their own facilities rather than leasing capacity may have somewhat more direct control over on-site sources than a typical public-chain validator does.

Scope 2 covers emissions embodied in purchased electricity and accounts for effectively the entire footprint. The point made in the preceding section applies here with equal force and in the same direction: where operators hold renewable supply contracts, their own corporate reporting may show a lower figure for the same infrastructure than this method produces, because this method uses the physical grid intensity of the supplying region rather than a contractual position. Neither figure is wrong; they answer different questions, and the one reported here is the grid-based one.

Greenhouse gas intensity per transaction is period emissions divided by transactions settled in the period, and carries the caveat given for energy intensity: the node population does not grow with usage, so the quotient describes an average across throughput rather than the emissions caused by one further transaction. Uncertainty compounds through the calculation, and its distribution is unusual for this network: the population is known with high confidence while the hardware and redundancy behind each operator is assumed, so errors originate later in the chain than they typically do. Figures are restated each period as operators disclose more. 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.