On 16 November 2022 the Australian Securities Exchange abandoned the blockchain system meant to replace CHESS, its clearing and settlement platform, and wrote off between A$245 million and A$255 million before tax. Thirteen days later Maersk said TradeLens, the shipping ledger it ran with IBM for more than 300 members, would shut down. Both were shared ledgers built for a closed circle of participants.
The public vs private blockchain argument has moved on since then. On 17 October 2026 the data site rwa.xyz counted $14.69 billion of tokenised US Treasury products held on open networks such as Ethereum, Solana and Stellar, with a BlackRock fund among the three largest. Behind the labels sit two plain questions: who may read the ledger, and who may add to it.
This lesson follows our lesson on what a crypto address is, where anyone could create a Bitcoin or Ethereum address without asking permission. A private blockchain takes that freedom away. For a refresher on blocks, hashes and nodes, read our guide to how a blockchain works first.
Search engines often mix this topic up with keys. A private key is the secret that controls one address, and the public key is the part you can share. Keys belong to a wallet; “public” and “private” blockchains describe a whole network. A private key works the same way on either kind of chain, and our guide to crypto wallets covers keys in detail.
Public vs private blockchain: who can read, and who can write
Most comparison pages sort blockchains into two boxes. The official texts ask two separate questions instead, and that alone clears up most of the confusion.
The first serious attempt to sort the vocabulary came from the US National Institute of Standards and Technology. Its report NIST IR 8202, “Blockchain Technology Overview”, published in October 2018, classifies blockchains by their permission model, meaning who may maintain the chain by publishing new blocks. If anyone can publish a block, the network is permissionless. If only users approved by some authority can, it is permissioned, and that authority may be one company or a group.
NIST then makes the point that trips up beginners. A permissioned network may still let anyone read the chain, or it may restrict reading to authorised people. Sending transactions works the same way: open to all, or only to approved parties. Reading and writing are separate switches.
European data protection regulators split the subject the same way. In their Guidelines 02/2025 on blockchain, adopted in final form (version 2.0) on 7 July 2026, the European Data Protection Board says the categories often break down into two elements: whether a chain is public or private, and whether it is permissioned or permissionless. The most common combinations, in the EDPB’s words, are public permissionless chains such as Bitcoin and Ethereum, and private permissioned chains, a “common choice within companies”.
So two questions do most of the work:
- Can anyone read the ledger? On a public chain, anyone with an internet connection can; on a private one, only members.
- Is adding blocks open to all? A permissionless chain lets anyone who follows the protocol’s rules validate; a permissioned one limits that job to participants an authority has approved.
Bitcoin answers “anyone” to both. A bank consortium ledger usually answers “members only” to both. The awkward cases sit in between, such as a network the whole world can read where a vetted group of validators adds the blocks. Keep both questions in mind and most marketing labels become easy to test.
Public, private, consortium and hybrid blockchains, with examples
With those two switches you can place any network in use today. The names below are the ones you will meet in the press and in project documents. Only the two axes come from official texts; “consortium” and “hybrid” are industry labels built on top of them.
Public permissionless: Bitcoin and Ethereum
Anyone can download the software, read every block since the first one, send a transaction and take part in validation. On Bitcoin that means mining with proof of work; on Ethereum it means locking ether as a validator under proof of stake. NIST notes that these networks usually reward the publishers of valid blocks with a native cryptocurrency, which is how they pay strangers to keep the ledger honest. Our guides to what Bitcoin is and how Ethereum works explain each network on its own.
Users appear only as addresses. Nobody checks your passport to send a transaction, although the exchange where you bought the coins almost certainly did, through its KYC checks.
Public permissioned: open to read, closed to validate
Here the ledger is visible to all, but only approved parties produce blocks. NIST explicitly allows for this mix. It is also where the arguments start, because a project can call itself “public” while keeping tight control of who validates. The Canton Network, covered further down, is the live example of that dispute.
Private blockchains: one operator decides
A single organisation controls who reads, who writes and who validates. NIST puts the trade-off plainly: if one entity controls who can publish blocks, users “will need to have trust in that entity”. A private chain can still make sense inside a group of subsidiaries, or as a tamper-evident log, but it does not remove the need to trust the operator.
Europe’s regulated market infrastructure offers a real case. CSD Prague, one of the first operators authorised under the EU’s DLT Pilot Regime, runs its settlement system on R3’s Corda Enterprise. According to a June 2025 report by ESMA, the EU markets authority, the system went live with a single validating node fully controlled by CSD Prague and later moved to two nodes.
Consortium blockchains: a club of companies
A consortium chain is permissioned, but the authority is a group of organisations rather than one. NIST observes that permissioned networks are often deployed for a group of organisations and individuals, “typically referred to as a consortium”, and that they suit firms that want to work together without fully trusting one another. Banks reconciling accounts with each other and shipping firms sharing documents are the textbook cases.
Much of the software is open source. On 16 September 2024 the Linux Foundation launched LF Decentralized Trust with 17 projects and more than 100 founding members. It now hosts Hyperledger Fabric, whose version 3.0 arrived under the new umbrella, and Besu, an Ethereum client used for private networks, which dropped the “Hyperledger” from its name.
Governments use the model too. In the European Blockchain Services Infrastructure (EBSI), every member of the European Blockchain Partnership, meaning the 27 EU countries, Norway, Liechtenstein and the European Commission, runs at least one node. Its first uses include notarising documents, verifying diplomas, cross-border digital identity and data sharing between tax and customs authorities. Running a node is reserved for partnership members.
Hybrid: a label, not a standard
“Hybrid” usually describes a permissioned service that lives on, or connects to, a public chain. No official text defines it, so treat it as a description to check rather than a category with rules.
The clearest regulated example is 21X, a German trading and settlement system authorised under the same EU pilot. ESMA’s report says it runs on Polygon PoS, a public, permissionless network, while access is restricted by smart contracts: only wallets that have passed KYC and anti-money-laundering checks and been approved by 21X can use it. Anyone can read the chain; very few can trade on the platform.
Difference between public and private blockchain, side by side
The table sums up the usual design choices. Each cell describes a typical network of that kind; real projects mix and match, so run the five questions at the end of this lesson before trusting a label.
| Question | Public permissionless | Private or consortium (permissioned) |
|---|---|---|
| Reading the ledger | Open to anyone | Usually members only, sometimes everyone |
| Sending transactions | Anyone who pays the fee | Approved members |
| Validating blocks | Anyone meeting the protocol’s rules (miners, stakers) | Named validators approved by an authority |
| Identity of participants | Pseudonymous addresses | Known organisations, identified before joining |
| Typical consensus | Proof of work, proof of stake | Raft, BFT protocols, notaries |
| Finality | Probabilistic on Bitcoin; about 15 minutes on Ethereum | Immediate in QBFT and IBFT 2.0 networks |
| Native token | Usually, to pay validators | Usually none |
| Changing the rules | Open proposals, client teams, node operators | The operator, the consortium or its foundation |
| Trust placed in | The protocol and its economic incentives | The operator or the members |
| Typical way it fails | Congestion, high fees, forks | Members stop paying or leave |
Nothing in the table is a score. A shipping consortium has no use for anonymous validators on the other side of the world, while someone sending stablecoins to a stranger does not want a bank able to refuse the payment.
Consensus: mining and staking versus known validators
Consensus is the method a network uses to agree on the next block. The lesson on how crypto transactions work followed one payment through that process on Bitcoin and Ethereum. Who is allowed to validate changes the method completely.
On a permissionless chain, one person could pretend to be a thousand validators. NIST explains that these networks therefore make block publishers spend or hold something scarce, computing power for proof of work or coins for proof of stake, so that cheating costs money. Bitcoin’s protocol aims for 2,016 blocks every two weeks, roughly one every 10 minutes, and its developer documentation says only someone with a majority of the network’s hashing power could reliably rewrite history. Finality there is probabilistic. Each new block on top makes a payment harder to reverse, but no single moment makes reversal impossible.
Ethereum works in 12-second slots grouped into epochs of 32 slots, and a block becomes final once validators holding two-thirds of the staked ether agree on checkpoints. According to ethereum.org, that takes about 15 minutes.
Permissioned networks know their validators by name, so they can use lighter methods. NIST says their consensus is “usually faster and less computationally expensive”, since identities are known and a member’s authorisation can be revoked if it misbehaves. Three families come up again and again:
- Hyperledger Fabric orders transactions with Raft, which tolerates crashes: a five-node ordering service keeps running if two nodes go down. Fabric 3.0 added SmartBFT, which also copes with nodes that behave maliciously, as long as fewer than a third of them fail.
- Besu networks use QBFT or IBFT 2.0, proof-of-authority protocols in which existing validators vote others in and out. Besu’s documentation describes immediate finality with no forks, asks for at least four validators to tolerate dishonest ones, and needs two-thirds of them online to produce blocks.
- R3’s Corda has no single central store of data. Each node keeps only the facts it is involved in, and a notary service stops double spends by refusing any transaction whose inputs have already been consumed.
Speed claims deserve care. The authors of the original Fabric research paper, posted in January 2018, reported more than 3,500 transactions per second with latency under a second “in certain popular deployment configurations”. That is a lab result from the people who built the software, not a measure of a live network. Real throughput depends on hardware and workload as much as on design, and no reliable, comparable figures exist across public and private chains.
The speed has a price. A network that finalises in a second because ten known banks sign each block is only as neutral as those banks. Public networks pay in time and energy so that nobody has to trust a named group.
Consortium blockchains from 2017 to 2026: what closed, what still runs
Between 2015 and 2018, large companies rushed to build “blockchain without the crypto”: permissioned ledgers shared by partners, with no public token. Several of the best-known ones are now gone, and their end dates are easy to check.
Four closures in nine months
we.trade went first. The trade-finance platform, a joint venture of 12 European banks and IBM built on Hyperledger Fabric, said in a memo dated 26 May 2022 that it had been “forced to discontinue” after failing to raise new money, GTR reported on 6 June. Its shareholders included Deutsche Bank, HSBC, Santander, Societe Generale and UniCredit.
The ASX project followed in November. The exchange had picked Digital Asset in 2017 to build its new settlement system. After an independent review by Accenture, chairman Damian Roche cited “significant technology, governance and delivery challenges”. The write-off of A$245 million to A$255 million before tax (A$172 million to A$179 million after tax) was booked in the first half of its 2023 financial year. The exchange added that the current CHESS remained “secure and stable” and was performing well.
TradeLens, launched by Maersk and IBM in 2018, came next. On 29 November 2022 Maersk said the platform had “not reached the level of commercial viability necessary to continue work”, and planned to take it offline by the end of March 2023. Rotem Hershko, Maersk’s head of business platforms, put it more bluntly: “the need for full global industry collaboration has not been achieved”.
Marco Polo closed the series. The trade-finance network, launched on R3’s Corda in September 2017 and backed at one stage by more than 30 banks, including Commerzbank, BNY Mellon and SMBC, entered insolvency in Ireland in February 2023. The Irish High Court appointed provisional liquidators after a planned $12 million deal with Bank of America fell through. Ledger Insights put its total debts at €5.2 million, with liabilities exceeding assets by €2.5 million.
In most of these cases the software did its job. The business model did not: rival firms had to join, share data and pay for a common platform run by someone else, and too few of them did.
What still runs: repo, reconciliation and central bank money
Permissioned ledgers did not disappear. They moved into the back offices of finance, where a small number of known institutions already deal with each other every day.
Broadridge, a US provider of back-office services, said on 9 April 2026 that its Distributed Ledger Repo platform handled an average of about $354 billion of repurchase agreements a day in March 2026, nearly $8 trillion over the month, a 392% rise in the daily average from a year earlier. Those figures are the company’s own.
In Italy, Spunta is the banking sector’s shared ledger for reconciling the accounts banks hold with each other. It is promoted by the Italian Banking Association (ABI) and coordinated by its innovation centre, ABI Lab, with R3’s Corda among its technical partners. In a release dated 5 September 2024, the Banque de France described Spunta as launched by ABI Lab in 2020 and handling 200 million transactions a year.
JPMorgan renamed its blockchain unit, Onyx, as Kinexys on 6 November 2024. At the time the bank said the platform handled more than $2 billion a day and had processed more than $1.5 trillion in notional value since launch, according to Ledger Insights. Like Broadridge’s, those numbers are self-reported.
Central banks are building on the same idea. The ECB said on 1 July 2025 that 64 participants had run more than 50 trials and experiments between May and November 2024, settling transactions recorded on distributed ledgers in central bank money. Three Eurosystem solutions were on test, among them Banca d’Italia’s TIPS Hash-Link and the Banque de France’s DL3S platform, which the French central bank also tested with ABI Lab and eight of its partner banks. In the same release the ECB set out a two-track plan: Pontes, to link such platforms to its TARGET settlement services, with a pilot due by the end of the third quarter of 2026, and Appia, a longer-term project. In a summary of decisions published on 2 October 2026, the ECB said its Market Infrastructure Board had recommended going ahead with the scheduled Pontes launch date of 21 September 2026.
Banks and fund managers moving to public blockchains
While consortia shrank, regulated institutions began issuing products on public chains, usually with controls added on top.
The European Investment Bank sold a €100 million two-year digital bond on Ethereum, priced on 27 April 2021 and settled the next day, with Goldman Sachs, Santander and Societe Generale as joint lead managers. Societe Generale’s digital-asset arm, SG-FORGE, acted as registrar, settlement agent and platform manager. Working with the Banque de France, the payment from the underwriters to the EIB was represented on the blockchain as central bank digital currency. The bond matured on 28 April 2023.
SG-FORGE then launched EURCV, a euro stablecoin, on Ethereum in April 2023 for the bank’s institutional clients. On 1 July 2024, the day after the EU’s MiCA rules for stablecoins began to apply, it restructured the token as an electronic money token and removed the whitelisting restrictions on transfers. On 20 February 2025 it announced the token’s arrival on Stellar. Our guide to what stablecoins are explains how such tokens are backed.
BlackRock launched BUIDL, which it called its first tokenised fund issued on a public blockchain, on Ethereum in March 2024, with Securitize as transfer agent. On its page dated 17 October 2026, read at about 14:00 UTC, rwa.xyz listed BUIDL at about $2.17 billion, the third-largest tokenised Treasury product behind Circle’s USYC and Ondo’s USDY. Across 117 products the site counted $14.69 billion of what it calls distributed value, meaning tokens held on public networks, spread across 86,754 holders and down 2.09% over 30 days. The networks listed include Ethereum, Solana, BNB Chain, Stellar and Arbitrum. These are data points, not recommendations.
JPMorgan picked Base, a layer 2 network on Ethereum incubated by Coinbase, for a pilot of its JPMD deposit token in June 2025. The Block, which described JPMD as a “permissioned” token, reported on 12 November 2025 that institutional clients could now send and receive it on Base. A bank-controlled product on a network anyone can read is the hybrid idea in its simplest form.
Canton and the fight over the word “public”
Some networks blur the line on purpose. Digital Asset’s Canton Network calls itself, on its home page, “the first privacy-enabled open blockchain network”. In a press release dated 17 December 2025 announcing work with DTCC, the main US securities settlement group, it went further, describing Canton as “the only public, permissionless blockchain purpose-built for institutional finance”. The release said DTCC had received SEC no-action relief for a service tokenising assets held at its depository, planned to mint a portion of the US Treasuries it holds on Canton, and would co-chair the Canton Foundation with Euroclear.
The Defiant does not accept the label. A newsletter signed by its founder argues that Canton is permissioned, because a would-be super validator needs a two-thirds vote of the existing ones. It also relays the view of Bartek Kiepuszewski of L2Beat that activity on Canton is self-reported and cannot be audited from outside, which makes it look more like a private database. A Canton representative, quoted in the same piece, says Canton shares the same permissionless properties as Ethereum and Solana.
Is a private blockchain just a database?
The doubt is as old as the idea. On 18 September 2015 the Princeton computer scientist Arvind Narayanan titled a blog post “Private blockchain” is just a confusing name for a shared database. Bruce Schneier, the security technologist, went further in “Blockchain and Trust”, an essay he posted on his blog on 12 February 2019: “Private blockchains are completely uninteresting.” To him they were nothing new, just shared append-only records with a list of approved writers. Elsewhere in the essay he warns that “Blockchain doesn’t eliminate the need to trust human institutions.”
NIST gives the other side. Its 2018 report says permissioned networks “can have the same traceability of digital assets” and the same distributed, redundant storage as open ones, and it notes that unlike traditional databases, past transactions on a blockchain are not overwritten. Auditors and supervisors can be given access to the ledger itself, “making audits a constant occurrence versus a periodic event”. For firms that know each other without fully trusting each other, that may be enough.
Regulators take a practical line. The EDPB’s checklist asks why a blockchain is necessary for the processing and what the alternatives are, and the board adds that an organisation which cannot use a permissioned chain should consider whether other technologies would suit it better. A permissioned ledger can be useful; its value comes from the members and the contracts between them, not from the word blockchain.
Privacy and the GDPR: why regulators lean towards permissioned chains
A public ledger shows every transaction to everyone, permanently. That is a strength for auditing a coin’s supply and a headache for personal data, which European law lets people have corrected or erased.
France’s data protection authority, the CNIL, tackled the question early, in guidance first published in French on 20 September 2018 (the English version is dated 29 October 2018). It said a participant who decides to record data on a chain “can be considered as a data controller”, that personal data should not be stored in cleartext on a blockchain, and that organisations should assess the “real necessity” of the technology, with a data protection impact assessment where needed. Private chains controlled by a single actor, it added, “do not raise specific issues regarding their compliance with the GDPR”, while public ones need particular attention on sub-contracting and transfers of data outside the EU.
The EDPB’s final guidelines of July 2026 go further. In general, they say, it is “not advisable to store personal data on the blockchain”, and the rights to erasure and to object “must be complied with by design”. On the choice of network, the board is direct: permissioned systems give a clearer allocation of responsibilities, so “organisations should favour permissioned blockchains”. Its checklist also asks whether a private chain would be enough, and whether a zero-knowledge architecture is possible.
For a user of Bitcoin or Ethereum, the practical point is simpler. Anything linked to your address stays readable by anyone, and privacy on a public chain depends on how you use addresses, not on the network hiding anything.
Rules in the EU, France, Italy, the UK and the US
No European rule bans or requires either type of chain. The laws regulate what is recorded and who offers the service, and the first licensed operators use both kinds.
The EU’s DLT Pilot Regime, Regulation (EU) 2022/858 of 30 May 2022, has applied since 23 March 2023. It lets licensed operators run trading and settlement systems for shares, bonds and fund units recorded on a distributed ledger, within limits: shares of companies worth less than €500 million, bonds under €1 billion per issue, funds under €500 million, no more than €6 billion of such instruments per infrastructure, and a transition plan once the total reaches €9 billion. In its report of 25 June 2025, ESMA counted only three authorised infrastructures as of 31 May 2025, CSD Prague, 21X and 360X, and suggested making the regime permanent with more flexible thresholds. The three illustrate the range: a private Corda network in Prague, a public Polygon chain with whitelisted wallets at 21X, and a private permissioned ledger run by Clearstream behind 360X.
MiCA, the EU law on crypto-assets, does not sort tokens by the kind of chain they live on. It regulates issuers and service providers; our guide to MiCA regulation covers who needs a licence, and the MiCA glossary entry gives the short version.
In France, beyond the CNIL guidance and the Banque de France’s DL3S tests, the AMF published a thematic file on applying the pilot regime on 11 December 2024, and on 9 April 2025 it put forward, with Italy’s Consob, proposals to make the regime more competitive.
Italy’s 2023 Fintech decree (decree-law no. 25 of 17 March 2023, converted by law no. 52 of 10 May 2023) allows certain financial instruments to be issued and transferred on a digital register. Consob keeps the list of authorised register managers under rules it adopted on 6 December 2023. In the Eurosystem trials Banca d’Italia supplied TIPS Hash-Link, while Spunta remains an initiative of the banks’ own association.
In the UK, the Bank of England and the FCA opened the Digital Securities Sandbox to applications on 30 September 2024, under regulations in force since 8 January 2024. It lets firms test the issuance, trading and settlement of securities such as shares, bonds and fund units on new technology; derivatives and unbacked cryptoassets such as bitcoin are out of scope. In the US, the SEC no-action relief granted to DTCC and cited in the Canton release is a recent case of a regulator letting a market infrastructure tokenise securities. It concerns a service, not a type of chain.
How to tell which kind of blockchain a project uses: five questions
Labels such as “public”, “enterprise-grade” or “hybrid” are not standard terms. Before you trust a token, a fund or a partner’s platform, put these five questions to it. The answers are usually in the project’s documentation or on its block explorer.
- Who runs the nodes? Look for a validator list. Thousands of unrelated operators point to a permissionless network; a handful of named companies points to a permissioned one.
- Could you run a validating node yourself without asking anyone? On Bitcoin and Ethereum the software is free to download and the protocol’s rules decide who validates. On a Besu or Fabric network, existing members admit new validators.
- Is there a public block explorer where you can look up a transaction without logging in? If you cannot see the ledger, it is private, whatever the brochure says.
- Does a native token pay the validators? NIST notes that permissionless chains usually reward block publishers this way. A token alone proves nothing, though: Canton has a coin and still needs a vote to admit super validators, so read this answer together with the first two. Our page on tokenomics explains what to look at in a token’s design.
- How are the rules changed, and by whom? Open improvement proposals and several independent client teams suggest a public network. A single company, a foundation or a membership agreement suggests a permissioned one, and also tells you who is legally responsible if something breaks.
Try it on Ethereum: thousands of independent validators, open staking, several public explorers, ether paying validators and upgrades debated in public proposals. All five answers point the same way. Try it on 21X or on JPMD and you get two answers at once, a public network carrying a service that one company controls.
The next lesson on this path will take common crypto myths one by one and check each against its primary source, the same habit these five questions train.
