How Algorand Users, Node Runners and Developers Can Back the Quantum Pitch

A pitch that needs evidence

William Herkelrath, Algorand's chief executive, argued on October 21 that banks cannot build settlement systems for decades of assets on cryptography a quantum computer could eventually break.¹ The post² is aimed at institutions weighing where to put tokenized deposits and other long-lived holdings.

A bank that takes the meeting will want evidence of three things before it commits. The first is post-quantum accounts used by ordinary people, not only by firms and test scripts. The second is a network run from many countries and many providers, one that keeps producing blocks. The third is people building on the chain and using what they build.

Each of those is something a reader can move, and the easiest of them is a few taps in a wallet. One thing to keep straight: Algorand's post-quantum work covers account keys and state proofs, while the consensus messages that produce blocks are still signed with Ed25519, as our special edition set out. That is exactly why evidence of real use counts for more than a feature list.

1. 2025-11-03 First post-quantum mainnet transaction, via logic signatures
2. 2026-06-18 Foundation publishes its post-quantum roadmap
3. 2026-08-13 Algorand v5.0.0 brings native Falcon-1024 accounts
4. 2026-09-17 Pera helps Algorand pass 1 million PQ transactions
5. 2026-10-21 CEO argues banks cannot build on breakable crypto

• 2025-11-03: First post-quantum mainnet transaction, via logic signatures.
• 2026-06-18: Foundation publishes its post-quantum roadmap.
• 2026-08-13: Algorand v5.0.0 brings native Falcon-1024 accounts.
• 2026-09-17: Pera helps Algorand pass 1 million PQ transactions.
• 2026-10-21: CEO argues banks cannot build on breakable crypto.

Open a quantum account, in a few taps

Pera Wallet is the one place where this is already a consumer feature, and it offers both routes. You can create a new quantum account from scratch, or upgrade an existing one and keep the same address and history.

For a new account:

1. Open Pera³, tap your account in the upper left corner, then the plus sign.⁴
2. Choose Add Quantum Account and name it.
3. Use it as you would any other account. Pera says an ASA minted from a quantum account counts toward its post-quantum transaction tally and comes out post-quantum resilient, with no extra setup.

For an account you already use, the same page³ offers an upgrade that keeps the address and its history. That upgrade is a rekey: Algorand's developer documentation⁵ describes it as the migration path, and says the public address stays the same while only the Falcon key can authorize its transactions afterwards.

Three things change, and each is worth knowing before you move anything valuable:

• The fee. The default minimum is 1,000⁶ microALGO; the documentation puts a Falcon-signed transaction's minimum at 3,000 microALGO (0.003⁷ ALGO).
• The tooling. The same documentation warns that wallets and apps may assume the 1,000⁸ microALGO minimum, or may not handle post-quantum signing at all, which can leave a migrated account unable to transact. Its own advice is to rekey when the threat is credible and your tools support it, so a small test account is the sensible first move.
• The door. A rekey is part of the account's ledger record. Close the account out and that record, including the authorized address, is deleted; fund the same address again and it is recreated under the original Ed25519 key.

Wallet support is uneven, and it is the honest reason to start small. Pera has shipped it. Lute⁹'s creator lists 'Falcon accounts (post-quantum, experimental)' among its features. Its published source appears to hold a dedicated Falcon signing path, taking the key type from the Algorand SDK and signing with a separate Falcon library. Defly¹⁰'s support is unconfirmed. Its manual covers multisig, Ledger accounts and trading, and lists no post-quantum feature, so whether it can sign for one today is unknown.

If an app rejects your quantum account, tell the app. Reports of a broken path are how a support matrix fills in, and the next reader benefits from the one you file.

One caveat about the scoreboard. Algorand crossed 1,000,000 post-quantum transactions within days of the feature going live. A significant share of that volume came from Pera's commemorative NFTs, handed out to anyone who connected a quantum account and sent a set number of transactions. The number shows the machinery works. It does not yet show that people are paying each other this way, which is the part a bank will look for.

Stake, with or without a machine

Running a node earns rewards only above a threshold. An account must stake at least 30,000¹¹ ALGO to be eligible. Algorand's staking page¹² says a community vote in Governance Period 10 set that figure. Below it, you can still hold stake online through a pool or a delegation service, and you can still be part of the number a bank checks.

That number is thinner than it looks. In the most recent consensus reading, about a fifth of staked ALGO, 20.2%, was online and participating. Every account that comes online visibly moves the share, and every one that leaves does too.

Valar¹³ is the delegation route. A holder keeps their keys and their ALGO in their own wallet, and a node runner operates the machine on their behalf under a service contract.

Réti is the pooling route. Stakers join pools run by validators, and Valar runs an interface¹⁴ for it. Read the staker-side terms on Réti's own pages before you deposit; the validator side is documented in more detail, and it tells you what to expect of an operator.

For a node runner who wants to serve other people's stake, both services publish their requirements. Valar's node guide¹⁵ asks for Python 3.12 or newer, then pip install valar_daemon. After that comes a configuration file holding four things: your validator ad id, the mnemonic of the Validator Manager hot wallet, and the URL and admin token for the algod instance on your machine. FUNC version 5 and later can run the same daemon from its menu. That mnemonic sits in a plain file on the node, so treat the machine as holding a live key and keep it patched.

Réti's validator quick start¹⁶ is more prescriptive about keys. It advises a hardware wallet or a multisig for the owner address, sets the manager address as a hot wallet, allows at most three pools per node and caps a validator at eight nodes.

PXke's operator runs a node that serves Valar delegators.

Run a node

This is the contribution that shows up most directly on the dashboard a bank would open. Algorand's run-a-node page¹⁷ asks for 16 GB of RAM, 8¹⁸ vCPU, a fast SSD of 100¹⁹ GB NVMe or equivalent, and a good connection, ideally 1²⁰ Gbps. Storage matters as much as memory: a cheap dedicated server with large slow disks does not qualify, however much RAM it carries.

With NodeKit

1. Install it on Linux or macOS with wget -qO- https://nodekit.run/install.sh | bash. If your Mac lacks bash, install it first with brew install bash.
2. Answer yes to the install prompt, then yes to fast-catchup. A regular sync takes days to weeks; fast-catchup takes 30 to 60²¹ minutes.
3. When the status turns green and reads RUNNING, press g to generate participation keys, enter your account address, and pick a duration. The documentation recommends 30 to 90²² days, and a 30-day key takes about four to six minutes to generate.
4. Press r to register the keys online. NodeKit assumes you want staking rewards and adds a fee at registration; start it with ./nodekit -n if you do not.
5. Sign the registration in a wallet: open the Lora link it shows, or press s for a QR code to scan in Pera.
6. Check that the account status reads ONLINE before you walk away.

With FUNC

1. Install with curl -fsSL https://raw.githubusercontent.com/GalaxyPay/func/main/install.sh | sudo sh.
2. Open http://localhost:3536 and choose the app password immediately. The service runs with administrator privileges, and until you set a password, anyone who can reach the app could set one for you.
3. Create the service and start the node, then wait for it to sync.
4. Connect a wallet, generate the participation key from the Participation Keys section, and register it.
5. Do not expose the app's ports to the internet. They are meant for machines on your own network.

What the machine costs

Prices below are list prices from the providers' own pages, and the cheap end of the market is no longer as cheap as it was.

Sources: 23 us.ovhcloud.com, 25 hostinger.com, 26 netcup.com, 28 hetzner.com

Hetzner used to be the default recommendation and is now the expensive answer, and on the captured page it could not be ordered at all. Hostinger's headline price roughly doubles at renewal, so read the term before you commit. OVHcloud and netcup also matter for the second reason a bank cares: placement.

Where the nodes are

According to Nodely's public dashboard³⁰, operators running fewer than six nodes each account for 21% over the last hour, with Amazon the largest named provider at 11%. The same reading puts Hetzner at 9%, OVH at 7%, Comcast at 6% and Verizon Business at 5%. Over the last 24 hours the dashboard puts the United States at 44%, with Germany a distant second and Ireland and France further back. Those are two different windows and should not be read as one snapshot.

According to the same dashboard, its full-time node estimate was 2,346³¹, down 18.3%³² over six months. The dashboard does not define what makes a node full-time, so treat the figure as a trend rather than a headcount.

Adding a machine in Brazil, Kenya, India or South Korea does more for the pitch than adding a fourth validator in Ashburn. Algorand's own best-practices page³³ asks cloud operators to watch the geographic and provider spread. It asks them to add to distribution rather than concentration. netcup runs data centers in Nuremberg, Vienna, Amsterdam, Manassas and Singapore. It calls Singapore its Asia-Pacific option, which makes a non-European location easy to buy from one provider.

Build, or make your app accept a quantum account

The hardest contribution is also the one that unblocks the others. A wallet or app that rejects a post-quantum account keeps its users off the chain, so every integration widens the usage proof.

The signing side is no longer undocumented at the library level. The JavaScript Algorand SDK's published source appears to contain a helper for it. That helper reportedly takes a Falcon-1024 public key and a function that signs bytes. According to the same source, it hands back the signers a transaction needs, plus an optional sending address for when the Falcon key authorizes a rekeyed account.

The fee is where careful code beats copied code. Algorand's fee documentation³⁴ sets the minimum as a base fee plus a contribution from the signature that authorizes the transaction, and a Falcon-1024 signature adds twice the base fee. Today that means 3,000 microALGO instead of 1,000, but the robust code reads the base fee from the network rather than writing either number in.

// Before: assumes every transaction costs 1,000 microALGO
const pay = await algorand.createTransaction.payment({
  sender: account.addr,
  receiver: bob,
  amount: algos(1),
  staticFee: microAlgos(1_000),
})

// After: the base fee comes from the network, the Falcon
// contribution (2x the base fee) is added, and a cap limits
// what congestion can cost
const sp = await algorand.getSuggestedParams()
const pay = await algorand.createTransaction.payment({
  sender: falconAccount.addr,
  receiver: bob,
  amount: algos(1),
  extraFee: microAlgos(Number(sp.minFee) * 2),
  maxFee: microAlgos(Number(sp.minFee) * 6),
})

For groups and app calls there is a second way. Since version 3.7.0 the JavaScript SDK's addressWithSignersFromRawPQSigner also returns an empty signer: simulate the group with it, and algod reports the fee a real Falcon signature would need, without producing one.

Size is the other difference. A Falcon-1024 signature is roughly 1.2³⁵ KB against 64³⁶ bytes for Ed25519, before you count the public key, so anything that batches transactions should count bytes as well as fees.

Why port now, when so few accounts use the feature? Because tooling gaps are the reason so few do. A developer who makes one wallet, marketplace or payment flow accept a quantum account removes one reason for the next reader to stay away. That developer also adds real transactions to the count a bank will ask about.

Start with the smallest one

The three proofs are not equally hard. Making a quantum account in Pera takes a minute, and using it for something ordinary takes another. Staking through a delegation service asks nothing of your hardware. A node costs about $25 to $50 a month and an evening of setup, or rather less if you already rent a server. Porting one flow to a 3,000 microALGO fee is a pull request.

Pick the smallest of those you can finish today. The pitch is being made in boardrooms, and the evidence it rests on is made, a little at a time, by people who have no meeting to attend.