Zcash’s new node Zakura goes live: privacy payments can reach 50k TPS, targeting Visa and Mastercard

Author: Coindesk

Compiled by: Deep Tide TechFlow

Deep Tide Insight: Zcash can currently process only about 1 private transaction per second. The goal of the new node Zakura is to reach 50k TPS while maintaining complete privacy—at the processing scale of Visa and Mastercard. This is an important signal for everyone focused on payment privacy and blockchain scalability: privacy and performance are no longer a trade-off.

Those rebuilding Zcash have a dream: to reach the level of global payment giants Visa and Mastercard—tens of thousands of payments per second—while maintaining full verifiability and strong privacy guarantees.

The first piece of this plan is Zakura, a new full node software that released version 1.0.0 on Wednesday. It is maintained by Sean Bowe, a founding member of Zcash’s zero-knowledge cryptography, and Dev Ojha, co-founder of Osmosis and the current head of Valar Group. Both teams are funded by private ZEC donations, not by companies or foundations.

“Our dream is to support global payments. Mastercard and Visa process more than 50k transactions per second; that’s our baseline. With Zcash’s current cryptographic scheme, that transaction volume requires node throughput of more than 500 MB per second,” a blog post said. “The existing tech stack can’t meet this target. The cryptographic technology our team is developing can close most of the gap.”

A full node is a program that stores a complete copy of the blockchain—here, the Zcash ledger—and independently checks every transaction to see whether it complies with the network rules. Zakura is a fork of the Zcash Foundation’s node software Zebra—meaning it was rebuilt from the Foundation’s official code.

Consensus rules are the shared rulebook executed by each node, determining which blocks and transactions the entire network accepts as valid. If one node applies different rules, it will fork and stop following the same chain as the other nodes.

Pruning, snapshots, and compatibility

Zakura can also prune, deleting old blockchain data that the node no longer needs, substantially reducing disk usage. This compresses the chain down to a small enough size. The team has released ready-made snapshots; after stripping old data, it’s about 11 GB, and new nodes can download it directly without pulling the full history piece by piece from other nodes.

This allows nodes to go from zero to running in less than two minutes, and the team says it’s “680 times faster.”

Compatibility mode further reproduces the interfaces of the original client zcashd, whose lifecycle will end on July 18, so wallet and exchange integrations built for it can continue to work as usual.

Throughput targets and Tachyon’s role

The reason for building all this is an arithmetic problem.

Mastercard and Visa process more than 50k transactions per second, and the team says this number is a “baseline, not a target.” Zcash’s current cryptographic scheme requires nodes to receive and verify more than 500 MB of data per second to keep up, because each private transaction carries a proof and the proofs are large.

This is roughly equivalent to continuously receiving a full DVD’s worth of data every 10 seconds, and no existing Zcash software can come close. The key is understanding why each bottleneck exists.

Bowe’s Tachyon project addresses this through recursive proofs: a single proof can prove the validity of thousands of other proofs, greatly reducing the amount of data that must be checked at the consensus layer.

Under the Tachyon design, nodes verify a single proof rather than thousands. The team says this will reduce consensus data demands from 100 MB per second to 500 MB, which they believe is technically achievable through careful engineering.

Wallet bottlenecks and Valar’s PIR solution

Wallets face a different problem. Because Zcash hides the receiver of a transaction, a wallet can’t ask a server which transactions belong to it without exposing itself. It must pull everything and test each item one by one—which is why wallet software can only reach about 1 transaction per second at most.

To eliminate this bottleneck, Valar Group is researching private information retrieval technology, so wallets can get their own data from the server without revealing which entries they’re requesting.

Fast block propagation

Fast block propagation means broadcasting newly mined blocks across the blockchain network as quickly as possible. Zakura is the software layer responsible for this task.

It must be fast enough to transfer new blocks between nodes so that large batches of proofs and wallet traffic can be supported. It comes with an experimental system designed to pass each block to every node within half a second, currently disabled by default.

Recent testing of these ideas will arrive in late July. Ironwood, officially named NU6.3, will be activated at mainnet block 3,428,143, around 8:00 AM Eastern on July 28, and Zakura has supported it since its release.

Origins of Ironwood

Ironwood exists because of a vulnerability that nearly destroyed Zcash in June.

The so-called “shielded pool” is the privacy layer of the network: both the amounts and participants are hidden, and zero-knowledge proofs serve as mathematical evidence.

On May 29, Shielded Labs researcher Taylor Hornby found that the proof circuit of Orchard—the latest shielded pool—contained a soundness bug that allowed an attacker to mint fake ZEC on-chain without leaving a trace. The bug had been present since Orchard’s activation in May 2022.

Developers disabled Orchard through an emergency response completed on June 2, and then restored it using the corrected circuit at block 3,364,600 via the NU6.2 hard fork on June 3.

The patch couldn’t fix the four years during which the vulnerability was open. Zero-knowledge proofs reveal nothing beyond the fact that verification passes, so there’s no on-chain record of any Orchard transaction transfers. No one can prove that fake ZEC was never created.

Ironwood was built to address this problem. The so-called “rotating door” at the boundary of the pool limits what can exit and enter, leveraging the fact that the ZEC amounts in and out of the shielded pool are public, even if transactions inside are private. By closing Orchard to stop new deposits, the rotating door becomes the only exit, and any fake coins inside get trapped there.

In simple terms, honest balances can leave over time, while fake coins may be prevented from fully exiting and circulating in the total supply. This traps any attempt to over-issue tokens at the boundary and restores reliability in how token supply is recovered.