Zcash developers shipped Zakura v1.0.0 on Wednesday, a node built around a single ambition: move private payments at the volume of a global card network. The project sets its floor at 50,000 transactions per second, the throughput CoinDesk reports Visa and Mastercard already clear. Zcash today settles roughly one private transaction per second.
The gap between those two numbers is the whole story. Closing it would turn a privacy coin that most people treat as a store of value into something that could plausibly clear day-to-day spending without exposing every payment on a public ledger.
The people and money behind Zakura
Zakura is maintained by Sean Bowe, a founding Zcash cryptographer, and Dev Ojha, an Osmosis cofounder who now leads the Valar Group. It is funded through private ZEC donations and runs independently of the Zcash Foundation, which maintains the existing Zebra node.
Zakura itself is a pruned fork of Zebra. The practical payoff is size and speed: the node uses about 11 GB of disk and syncs in under two minutes, which the developers frame as roughly 680 times faster than a traditional full sync. It also ships a compatibility mode that reproduces the interface of zcashd, the legacy node that reached end-of-life on July 18. That matters because wallets and services wired to the old software need somewhere to land.
The throughput math
Getting from one transaction per second to 50,000 is not a matter of raising a block-size constant. Verifying that volume of shielded activity means processing on the order of 500 MB per second of data, and doing it while keeping sender, receiver, and amount hidden.
Two research efforts carry that weight. The first is Project Tachyon, Bowe's work on recursive zero-knowledge proofs, which compresses the consensus data a node has to check rather than forcing every participant to replay all of it. The second is private information retrieval research from Valar, which lets a wallet fetch its own transaction data from a server without revealing which records it asked for. Bandwidth and lookup privacy are the two bottlenecks that have historically kept anonymous payments slow, and each piece targets one of them.
None of this is shipped at 50,000 TPS today. The 50,000 figure is a design floor the team is building toward, not a benchmark Zakura hits now. Treat the throughput claim as a roadmap, not a current spec.
Ironwood and a bug fix arriving July 28
Zakura lands days ahead of a scheduled network upgrade. Ironwood, formally NU6.3, activates on July 28 at block 3,428,143, and Zakura supports it from release. Ironwood introduces a "turnstile" mechanism that caps withdrawals from the Orchard shielded pool. That control responds to a soundness bug discovered on May 29, 2026, one the developers say had been present since May 2022. Capping outflows limits the damage any exploit of that flaw could cause while the fix propagates.
The sequencing is deliberate. A node built for scale is worth little if the shielded pool underneath it has an open soundness question, so the throughput ambition and the safety patch are shipping in the same window.
The payments framing, and its limits
Pitching a privacy chain against Visa is a statement of intent more than a competitive claim. Card networks settle through banks, chargebacks, and fraud desks; a shielded chain settles with cryptographic proofs and no built-in reversal. The comparison the team is really making is about capacity, not about replacing the rails a cardholder swipes through today.
Still, throughput is the precondition for any of it. Privacy-preserving payments have stayed niche partly because confidential transactions are computationally heavy and slow to verify, which is exactly the pressure regulators lean on when they treat anonymity tooling as suspect. Chinese prosecutors, for instance, have pushed to presume money laundering from mixer use. A network that can process private payments at card-network speed does not resolve that tension, but it changes what a privacy chain is technically capable of carrying.
Overview
Zcash's Zakura node, released July 16, 2026, by Sean Bowe and Dev Ojha, sets a 50,000 TPS floor meant to match Visa and Mastercard, against Zcash's current pace of about one private transaction per second. It syncs in under two minutes on roughly 11 GB, replaces the retired zcashd interface, and supports the Ironwood upgrade activating July 28, which caps Orchard pool withdrawals to contain a soundness bug found in May. The headline number is a target, not a benchmark, and the work that could deliver it, recursive proofs and private information retrieval, is still research.
