Algorand's quantum pitch: the keys are ready, the chain is not yet

William Herkelrath has spent his first two months as chief executive of the Algorand Foundation arguing that banks should choose a blockchain by its cryptography. In an interview published on October 22¹ he said he expects deals with major banks in 2027. The day before, in a blog post², he wrote that a financial system built for the future 'cannot depend indefinitely on cryptographic assumptions that future technology may invalidate.³'

That is the clearest statement yet of where the Foundation is going: fewer retail gestures, more paid work for institutions, with post-quantum security as the product. What makes it worth reading closely today is that Algorand's own documents, and its own chain, show precisely which half of the promise is already true, and which half a bank would be buying on trust.

The plan is a revenue plan

Herkelrath has been unusually direct about the motive. Asked whether he anticipated deals with major banks in 2027, he said: 'I think so. Yeah, absolutely. Can't announce anything,⁴' adding that conversations were 'very positive.' In a September AMA on Reddit he wrote that if Algorand 'is being used to provide a useful service that people are willing to pay for⁵', the Foundation could 'use that new revenue to become net buyers of the token.⁶' He has called self-sustainability 'the primary focus for Algorand 2.0' and told the same AMA that enterprises want services more than infrastructure.

The timing was not incidental. The US Office of Management and Budget's memorandum M-26-15⁷, issued on June 24, 2026, gave federal agencies 120⁸ days to submit a post-quantum migration plan. That deadline fell on October 22. The memorandum set the objective of mitigating as much quantum risk as feasible by December 31, 2030. It does not apply to national security systems, and it tells agencies to start with high-value assets and the systems most likely to be vulnerable. Those are the customers Herkelrath is describing. He had been announced to discuss exactly that deadline on a podcast on October 22; he wrote on X that the appearance had been rescheduled, and gave no new date.

1. 2026-06-18 Foundation publishes post-quantum roadmap, targets 2027
2. 2026-08-20 v5.0.0 ships native Falcon-1024 accounts
3. 2026-09-09 Herkelrath announced as CEO, effective Aug 31
4. 2026-10-09 CTO Bruno Martins resigns
5. 2026-10-14 Strategy and marketing chief resigns
6. 2026-10-22 Treasury consolidated; bank-facing interview airs

• 2026-06-18: Foundation publishes post-quantum roadmap, targets 2027.
• 2026-08-20: v5.0.0 ships native Falcon-1024 accounts.
• 2026-09-09: Herkelrath announced as CEO, effective Aug 31.
• 2026-10-09: CTO Bruno Martins resigns.
• 2026-10-14: Strategy and marketing chief resigns.
• 2026-10-22: Treasury consolidated; bank-facing interview airs.

Two items in the record show where the pitch sits against the market it is aimed at. The one documented bank engagement on Algorand that concerns real deposits is not a quantum sale. VersaBank's June 2026 SEC filing named Ethereum, Algorand and Stellar for tokenized deposits that represent the bank's own liabilities rather than a stablecoin. And the post-quantum custody pilot that banks and regulators actually joined this year is being tested on a NEAR testnet. It is run by the Responsible Fintech Institute and Safeheron, with Bison Bank and DK Bank taking part and regulators from Abu Dhabi, Bhutan and Malta observing. On bank-facing quantum custody, another chain moved first.

What the keys protect, and what they don't

The part of the pitch that protects a holder's own keys is real, shipped and rare. Algorand's developer documentation⁹ describes post-quantum accounts that hold Algos and assets, send transactions and call applications like any other account. They are authorized by a Falcon-1024 signature instead of an elliptic-curve one. The chain's own account records make the difference visible: a wallet created with a Falcon key reads as signature type 'pqsig', while an ordinary address reads 'sig'. In the first nineteen days after the feature went live in August, the Foundation counted 1,800 such accounts and 1.24 million¹⁰ transactions signed from them.

For a holder deciding what to do with that, the same documentation is candid about the trade-off. Rekeying keeps the address and moves control to the Falcon key, but each transaction then carries a higher minimum fee, and not every wallet or application handles post-quantum signing yet. Its advice is to migrate when the threat is credible and the tooling supports it, which is a fair summary of where the feature stands.

The rest of the chain has not made that move, and the Foundation says so. Its June roadmap¹¹ states that the verifiable random function behind consensus 'relies on elliptic-curve cryptography and thus is not fully resistant to quantum adversaries', and that consensus messages are signed with Ed25519, 'which is vulnerable to quantum attacks.¹²' The documentation adds that there is 'No post-quantum native multisig. Multisig accounts remain Ed25519-only by design,' and that consensus participation is unchanged. A post-quantum random function is the subject of a research paper targeted for early 2027; consensus signing is still being researched.

Google's researchers made the same point more bluntly, in a paper the Foundation reprinted in full on its own blog. 'Algorand's consensus and builtin transactions are based on quantum-vulnerable Ed25519 digital signature scheme,¹³' the March 2026 whitepaper says, while crediting the network with deploying Falcon signatures for smart transactions and state proofs. Read together, the documents describe a chain whose account keys are quantum-resistant and whose block production is not.

In the interview Herkelrath went further than those documents do, describing the accounts on Algorand as already quantum resistant, and he put full readiness at the end of 2027. For a holder who creates or rekeys to a Falcon key, the first claim is accurate. For the Foundation itself, it was not accurate the same week.

The test the Foundation set itself

On the afternoon of October 22, five wallets on the Foundation's published treasury list sent about 193.6 million¹⁴ ALGO into one new address. The money moved in five payments between 14:32:35 and 14:33:08 UTC. The destination had been created that morning by a 1 ALGO payment from one of those same addresses. It held 193,589,511.6¹⁵ ALGO afterwards. Two of the five senders were left holding 0.2¹⁶ ALGO, their minimum balance. Nothing left the Foundation: this was an internal consolidation, and the payments are recorded as received in the destination's own history.

What a reader could not have known that day is where the money went. The destination was not on the Foundation's published list¹⁷ of treasury addresses. A community member asked about it on Reddit the next morning, and a Foundation staff account replied that it 'Might have been a miscommunication between finance and marketing on the publishing¹⁸', then named four addresses to be added; the page now carries an update stamp of October 23.

That is a publishing lapse, quickly answered. The finding is what the destination is. The account now holding about a fifth of the Foundation's reported 1.04 billion¹⁹ ALGO at the end of June is an ordinary single-signature address, as the ledger reads it. In June the Foundation said it would 'begin migrating its own treasury to post-quantum accounts²⁰' later in 2026, and demonstrate post-quantum multi-signature custody while native support was still being built. That window has not closed. But the Foundation set a public, checkable test for itself, and the chain says it has not been met yet.

Rewards outrun fees, and that is the point

The Foundation's Q2 transparency report is the last full look at its balance sheet, and it states that the figures are unaudited. It put holdings at 1,035,603,000 ALGO on June 30, down 43,272,000 in the quarter. At the $0.1075 ALGO traded at on October 23, that is about $111 million. The Foundation sold 10.60 million²¹ ALGO through structured selling in those three months and spent 2,788,000 ALGO on marketing, events and partnerships. Its table pairs each ALGO figure with a dollar figure for the same quarter, so the columns are one spend in two units, not two periods.

The line that explains the strategy is the reward bill. Validators were paid 20,124,000²² ALGO in the second quarter, and 6.74 million²³ ALGO in August alone, taking distributions for 2026 to 53.63 million by the end of that month. Fees are nowhere near. The chain processed about 1.16 million²⁴ transactions a day over 24 hours and a similar average over 30 days. At the 0.001²⁵ ALGO minimum fee that is roughly 35,000 ALGO a month: a floor rather than a measurement, since many transactions pay more, and still under one percent of August's reward payout.

The Foundation described this transition when staking rewards launched. 'At first, rewards will come from the Algorand Foundation and network fees,²⁶' its guide said in 2024, 'but, over time, as adoption grows and fee structures change, the system will transition toward self-sustainability.²⁷' The bank pitch is an attempt to bring the second half of that sentence forward. It is a revenue question rather than a cryptography question, and the runway is long enough to answer it: on the quarter's net movement the treasury lasts several years, and the Foundation's summary of the September AMA said the chief executive believes self-sustainability is reachable in under five.

What the traction is worth

The strategy's early results are real and small, at the size its design expects. The Foundation's x402 challenge²⁸, which asks developers to sell API calls to AI agents, drew 115²⁹ projects that put paid endpoints on mainnet; 20 shared a 500,000³⁰ ALGO leaderboard pool and 10 pitch for a $100,000³¹ prize on October 30. Settled volume in August was $174,000³² across 615,000³³ transfers, about 28 cents each.

Post-quantum usage has the same shape and a sharper caveat. The 1.24 million³⁴ transactions in nineteen days led the Foundation's August report to describe 'meaningful usage intensity among early adopters.³⁵' A large share of the recent count, though, comes from a free web tool³⁶ that generates a Falcon account in a browser tab and prints post-quantum transactions until its fee balance runs out. Its application's stored counter reads about 1.65 million. Those transactions are signed, settled and paid for, so nothing about them is fake. What they demonstrate is a capability that a handful of browsers can drive, not demand from institutions. We looked at what made up the headline count earlier this month.

Resignations, not layoffs

The week's other news looked from a distance like a staff cut. On the record, it was resignations. Bruno Martins, the chief technology officer, announced on October 9³⁷ that after nearly four years he had decided to 'resign and move on', and gave no reason.³⁸ The chief strategy and marketing officer resigned on October 14, and a community post on October 22 said the head of product marketing had been let go. The Foundation's only announced workforce reduction remains the 25%³⁹ cut it made in March 2026, which it attributed to the macro environment and the crypto downturn. No October reduction, and no size for one, has been published.

A vacant seat matters less here than it would at a company that ships the protocol single-handed. The Foundation's team page⁴⁰ still lists Chris Peikert as chief scientific officer and five other executives, and the cryptography has a named owner. What the departures thin out is the retail-facing side of the organization, which was already the side the new strategy addresses least.

The community split along that line. One long-time member wrote that he did not think the change was necessarily bad for the token, and that a lot of staff needed to go. Others doubted the premise, noting that selling to institutions had been tried for years without much to show. Both readings fit the record. What is not in dispute is that the audience the Foundation addressed most directly for six years, retail holders, collectors and small builders, is no longer the one its chief executive is speaking to.

What to watch

Two figures describe the ground the pitch stands on. About 20% of staked ALGO was online on October 23, 1.98 billion of a total stake of 9.81 billion, which is the share of supply actually securing the chain a bank would be joining. And the site's FAQ still describes transaction fees as accumulating in a dedicated account while the community discusses what to do with them, a description that no longer matches what the account does.

Neither figure undercuts the strategy. Together they suggest that the Foundation is selling a capability ahead of its market, which is a normal thing for a new product to do, to institutions whose questions are narrower than the slogan. A bank's reviewer will ask what the network guarantees today, not what it guarantees in 2027.

The strategy is not a retreat, and the record does not support reading it as one. Algorand shipped State Proofs signed with Falcon in 2022 and native post-quantum accounts in August, and Google's researchers cite the work. The distance between that and the claim a bank is being asked to buy is measured in dates the Foundation chose for itself. Post-quantum multi-signature custody and the start of the treasury migration were both promised for later in 2026. Native multi-signature is Ed25519-only today. The next protocol release and the next transparency report will show whether the promise is being kept on schedule. Until they do, the fairest description of Algorand's quantum position is the one its own documents give: the keys are ready, the chain is not yet.