Zcash processes roughly one private transaction per second today. The team behind a new node client released Wednesday wants to push that number past 50,000, matching the throughput that Visa and Mastercard handle during peak shopping seasons.
Zakura 1.0.0, the first independently maintained Zcash full node outside the Zcash Foundation’s control, marks the opening move in an ambitious multi-year scaling effort. The project is led by Sean Bowe, one of the original architects of Zcash’s zero-knowledge cryptography, and Dev Ojha, who co-founded the Osmosis decentralized exchange before starting Valar Group. Neither team takes foundation or company money; both run on private ZEC donations.
Why Current Zcash Cannot Handle Payment-Network Volume
The arithmetic problem is brutal. Every shielded Zcash transaction carries a cryptographic proof that the sender owns the coins being spent and that no new coins are being created from thin air. These proofs are large. At Visa-scale throughput, a node would need to ingest and verify more than 500 megabytes of proof data every second, the equivalent of a full DVD arriving every ten seconds, continuously, around the clock.
No existing Zcash software comes close to that. The legacy zcashd client, which reached end of life on July 18, struggled with far lighter loads. Zebra, the Zcash Foundation’s replacement node, improved on zcashd but still operates orders of magnitude below payment-network requirements.
Zakura does not solve the throughput problem by itself. What it does is lay the foundation for two other research efforts, Project Tachyon and Valar’s private information retrieval work, that attack the bottlenecks from different angles.
Zakura’s Technical Architecture: Pruning and Snapshots
Zakura forked from Zebra’s codebase and rebuilt it with aggressive disk optimization. The node can prune, meaning it deletes old blockchain data no longer required to validate new blocks. That shrinks storage requirements dramatically.
The team publishes pre-built snapshots of the pruned chain, about 11 gigabytes with historical data stripped out. A new node operator can download this snapshot and start validating in under two minutes. The team claims this is 680 times faster than syncing block-by-block from peers.
For operators still running infrastructure built against zcashd, Zakura includes a compatibility layer that reproduces the old client’s interface. Exchange integrations and wallet backends that communicated with zcashd will continue working without code changes.
Project Tachyon and Recursive Proofs
The raw throughput gains will come from Project Tachyon, Bowe’s research initiative into recursive zero-knowledge proofs. The concept sounds abstract but the mechanics are straightforward: instead of a node checking thousands of individual transaction proofs, it checks a single proof that attests to the validity of all the others.
Think of it like an auditor reviewing a company’s books. Checking every receipt from every employee expense report would take weeks. But if a trusted accountant has already verified each report and signed a summary statement, the auditor only needs to verify that one signature. Recursive proofs work similarly, with cryptographic guarantees replacing the need to trust the accountant.
Under current Zcash cryptography, consensus-level data requirements run around 100 megabytes per second for moderate transaction volumes. Tachyon aims to compress that to 500 megabytes per second at full Visa-scale load. The team believes that figure is achievable with careful engineering, though no timeline has been published.
The Wallet Bottleneck and Private Information Retrieval
Nodes are only half the problem. Wallets face their own constraint, and it stems directly from Zcash’s privacy model.
Because shielded transactions hide the recipient, a wallet cannot simply ask a server for transactions addressed to it. Doing so would reveal which addresses the wallet controls, defeating the privacy guarantees. Instead, Zcash wallets download every transaction on the network and test each one locally to see if it belongs to them. This trial decryption process caps current wallet software at about one transaction per second.
Valar Group is attacking this bottleneck with private information retrieval, a cryptographic technique that lets a client fetch specific database entries without the server learning which entries were requested. If the research succeeds, a wallet could query a server for its own transactions while the server remains mathematically unable to determine which data was retrieved.
This is particularly relevant for mobile wallets and light clients. Right now, running a Zcash wallet on a phone means either sacrificing privacy by using a trusted server or waiting hours for the wallet to scan the entire blockchain locally. Private information retrieval could eliminate that tradeoff.
Ironwood Upgrade Addresses the Soundness Bug
Zakura ships with support for Ironwood (NU6.3), a network upgrade activating July 28. The upgrade introduces a turnstile mechanism that caps the rate at which ZEC can be withdrawn from the Orchard shielded pool.
The reason for this measure traces back to a critical vulnerability disclosed in June. A soundness bug in Zcash’s Orchard pool raised the possibility that counterfeit ZEC had been secretly created. Because Zcash balances are encrypted, there is no straightforward way to audit whether such counterfeiting occurred.
Ironwood’s turnstile does not fix the underlying bug (that required a separate cryptographic patch) but does contain potential damage. If counterfeit coins exist in Orchard, the withdrawal cap prevents an attacker from rapidly draining them into the transparent pool where they could be sold.
The Ironwood testnet deployment reported no serious bugs during security audits, giving operators confidence that the mainnet activation will proceed smoothly.
What 50,000 TPS Would Mean for Privacy Cryptocurrency
The throughput target is not arbitrary. Fifty thousand transactions per second represents the approximate peak capacity of Visa’s network. Mastercard operates at similar scale. Any cryptocurrency aiming to displace card networks for everyday payments needs to match those numbers.
Bitcoin handles roughly seven transactions per second on its base layer. Ethereum manages around 15 to 30, depending on gas prices and block utilization. Layer 2 networks like Arbitrum and Optimism push throughput higher but inherit the transparency of their base chains.
Zcash’s challenge is achieving scale while preserving privacy, a combination no major network has demonstrated in production. The cryptographic overhead of zero-knowledge proofs makes every transaction more expensive to verify than a simple balance transfer.
If Bowe and Ojha’s teams deliver, Zcash would occupy a unique position: a privacy-preserving network capable of handling the transaction volume of a global payments processor. The business implications extend beyond retail payments. Institutional treasury operations, supply chain settlements, and healthcare data transfers all involve information that parties prefer not to broadcast publicly.
The engineering hurdles remain substantial. Recursive proofs at scale have never been deployed in a production blockchain. Private information retrieval exists in academic literature but lacks real-world implementations handling millions of users. Whether Zakura becomes the foundation for a payments revolution or remains a fast-syncing node for enthusiasts depends on research breakthroughs that have not yet occurred.
Zakura 1.0.0 is available now. What comes next requires years of cryptographic work that no one has finished.
