Zcash is built by several teams. Work on network upgrades, sustainability, new assets, finality and scalability each moves at its own pace. This page collects what is already live and what is still in preparation.
Where things stand
2026 was an unusually busy year for Zcash. An Orchard vulnerability found in late May was fixed with an emergency upgrade in early June, and in late July Ironwood opened, a new shielded pool whose supply can be verified again. In November, NU7 activated, shortening the block interval and changing the fee structure. The ‘State’ column in the table below refers to mainnet.
Recovery path applied to Ironwood; Epoch targeted for late 2027
Changes to the Zcash protocol are proposed as ZIPs (Zcash Improvement Proposals) and activated together in network upgrades (NU). What goes in is decided through community panels and coinholder sentiment polls. How decisions are made →
Recent network upgrades
NU5: Orchard
The Orchard pool was introduced, using the Halo 2 proof system, which needs no trusted setup, and unified addresses (UA).
NU6: Dev Fund restructuring
Part of the block reward began to be held in a lockbox instead of being spent right away. Details →
NU6.1: Coinholder-Controlled Fund
Funds in the lockbox, plus 12% going forward, were turned into a fund whose use coinholders decide by vote (ZIP 1016). This structure runs until around November 2028. Details →
Emergency response to the Orchard vulnerability (NU6.2)
Independent researcher Taylor Hornby, who was auditing for Shielded Labs, found an Orchard forgery vulnerability and reported it privately. In response, the Orchard pool was briefly halted by a soft fork, and NU6.2 applied the remaining fixes. Details →
NU6.3: Ironwood
Activated at height 3,428,143. Orchard became spend-only, and new funds go into the Ironwood pool. Ironwood notes are built to be quantum-recoverable from the start. Details →
NU7
Activated at height 3,543,405 (02:34 UTC, November 5 in US time). The block interval dropped to 25 seconds, 60% of fees were removed from circulation, and Sprout spends were blocked. Details →
NU8
No date has been set. A new transaction format (ZIP 248) and a burn field (ZIP 233) have been mentioned as candidates.
The history of the shielded pools since Sprout in 2016 is on the Privacy page.
NU7: what changed
NU7 activated on November 6, 2026 at 02:34 UTC (11:34 KST; November 5 in US time zones) at mainnet height 3,543,405. The testnet went live first, on October 6. Its scope was set through the community panel (ZCAP) and a coinholder sentiment poll in August and September.
ZIP 218
Block interval 75 seconds → 25 seconds
Blocks come three times as often, but the block reward is one third and the halving interval (in blocks) is three times longer. Daily issuance (about 1,800 ZEC) and the halving timing stay the same. Throughput for Orchard-style protocols (now Ironwood) rises from about 2.9 to about 6.6 transactions per second. Read the ZIP →
ZIP 235
Remove 60% of fees from circulation
60% of transaction fees go to a separate reserve instead of to miners. Miners receive 40%. This is the first step of NSM. Read the ZIP →
ZIP 237
Re-issuance that preserves the halving
A rule for releasing the ZEC collected in the reserve as future block rewards. Re-issuance starts at mainnet height 8,939,664 (around February 2031). Read the ZIP →
ZIP 2003
Ban v4 transactions (Sprout freeze)
Because v4 transactions are no longer accepted, funds in the Sprout pool can no longer be spent. The first shielded pool from 2016 is effectively closed. Read the ZIP →
ZIP 2008
ZCG funding address update
The list of addresses that receive the ZCG (community grants) 8% share of the block reward was reset for operational reasons. Read the ZIP →
Zcash’s mining reward halves every four years, and at some point fees alone will have to cover the cost of securing the network. NSM is a mechanism proposed by Shielded Labs that removes part of the fees from circulation, accumulates them in a reserve, and releases them again as block rewards in the distant future to build a long-term security budget. The total supply cap of 21 million stays the same.
What went into NU7
Removing 60% of fees (ZIP 235) and a re-issuance rule that preserves the halving (ZIP 237). To avoid changing the transaction format, a new value pool was added, and the removed fees and unclaimed mining rewards collect there. It is Shielded Labs’ first mainnet contribution.
What the sentiment poll decided
In the September 2026 sentiment poll, the option that preserves the halving was chosen over the option that flattens the issuance curve. On when re-issuance should start, the panel (as soon as possible) and coinholders (February 2031) disagreed, and the more conservative February 2031 was chosen. Call #132 →
Next candidate
The burn field (ZIP 233), which would let anyone voluntarily send ZEC to the reserve, needs a transaction format change, so it moved to being an NU8 candidate.
This feature lets assets other than ZEC (stablecoins, tokens and so on) be issued and transferred on the Zcash shielded pool. The asset type is hidden too, so which asset was moved is not revealed either. It has been developed by QEDIT, and the relevant ZIPs are 226 (transfer and burn), 227 (issuance) and 228 (asset swaps). All are in Draft status.
Feb-Mar 2026
Mainnet put on hold
In the February 2026 NU7 sentiment poll, 70.4% of the community panel (ZCAP) supported it but 98.6% of coinholders opposed it. So the ZSA code was placed behind a feature flag, with the plan to verify its usefulness on the testnet and then decide on mainnet activation.
Aug-Oct 2026
Dedicated testnet and Ironwood support
A dedicated public ZSA testnet opened in August, and in October the work of catching up with the Ironwood (NU6.3) changes was completed. A handover of the code to the Zcash repository is also under discussion.
Crosslink: proof of work + proof-of-stake finality
Today’s Zcash, like Bitcoin, produces blocks with proof of work (PoW), and transactions are only probabilistically final. Crosslink is a design that adds a proof-of-stake (PoS) based BFT finality layer on top of the PoW chain to provide irreversible final settlement. ZEC holders would be able to earn rewards by staking, and the 21 million cap would be kept. It is being developed by Shielded Labs.
Mar 2025 - Apr 2026
Design and prototype phase completed
Milestones 1 to 5 of the public roadmap are done, and a phase of refining it to production quality is underway from Q2 2026 to Q2 2027.
Oct 2026
Incentivized testnet
A new FeatureNet restarted on October 1, and testnet participants (mining, staking and contributing) are being paid ZEC rewards. The team said feature implementation is done but that further design review is needed before it can go into a network upgrade.
No mainnet activation date or target upgrade has been set.
Project Tachyon: scalability and post-quantum privacy
Led by Sean Bowe, a co-designer of Halo, this project redesigns the protocol so it can handle far more shielded transactions than today. The official introduction lists reducing transaction size to about one hundredth and eliminating the problem of validators carrying ever-growing state as its goals. It is still in the design and development phase, so there are no measured figures.
Why scaling is hard: four bottlenecks
Scaling shielded payments cannot be reduced to a single TPS number. Wallets, blocks and node state each get stuck in a different place. On top of that, Zcash makes fully shielded transactions indistinguishable from one another, so common scaling techniques used by other chains cannot simply be reused.
Wallet
Creating proofs
Since Sapling in 2018, shielded transactions can be created even on phones, and Orchard removed the trusted setup as well. This is the bottleneck that has been solved the most of the four.
Wallet
Finding your money (syncing)
Shielded outputs carry no address, so a wallet has to try opening every output with its own keys. Syncing gets slower as usage grows. Trial decryption →
Block
Block space and verification
Shielded transactions are large and expensive to verify. With NU7’s 25-second blocks and the Action limit set alongside, the maximum throughput of Orchard-style transactions (assuming 2 Actions) rose from about 2.9 to about 6.6 transactions per second, but that is a parameter adjustment with the structure unchanged. Ironwood is also a new pool for security, not a capacity upgrade.
Node
Accumulating state
To prevent double spending, every node must keep the published nullifiers forever. According to the Tachyon team, this can grow by several GB per day even at moderate throughput.
Tachyon’s approach
The key tool is PCD (proof-carrying data), which chains proofs together. Each new proof absorbs what was verified in the previous step, so a long history does not have to be re-verified every time.
Block space · verification
Block producers bundle proofs
Each wallet proves only its own transaction, and the block producer later folds the finished proofs into one. Wallets do not need to coordinate, the size and verification cost per transaction drop, and the bundling work can scale horizontally.
Node state
Oblivious syncing
A remote service updates the proof that ‘this money has not been spent yet’ on the user’s behalf without knowing the contents (obliviously). Nullifiers are periodically changed so they cannot be linked, so the service cannot learn the real nullifier. This puts a cap on the state validators must hold.
Syncing · quantum
Separating the payment protocol
Payment information exchanged between wallets is split off from the on-chain shielded protocol. This removes quantum-vulnerable privacy assumptions from the chain and opens a path to post-quantum privacy.
Compatibility
On top of today’s ecosystem
It needs a new shielded pool that uses PCD from end to end and a simpler key structure, but it is being designed to stay compatible with existing cryptography such as hardware wallets.
Progress
Ragu
A Rust library for Halo-style PCD. It is in development, with optimization and audits still to come. Since September 2026, the mature parts have been moving into Zakura’s shared crates.
Payment protocol
Approaches using PIR (private information retrieval) and post-quantum key exchange are being explored together.
New shielded protocol → testnet → consensus and ZIP → mainnet
The official roadmap gives no dates.
The Tachyon team, together with Valar Group, also built the formal verification of Ironwood, the Zakura full node and the coinholder voting chain.
zcashd, the original node used since 2016, does not support Ironwood, and its support ended just before the July 2026 activation. Nodes are now handled by the Zcash Foundation’s Zebra and Valar Group’s Zakura, and wallet RPC by Zallet and Zecd.
Node
Zebra 7.0.0
A stable release with NU7 support (Oct 22, 2026). It fixed 7 security advisories. A guide for migrating from the zcashd RPC is published in the Zebra book.
Node
Zakura
A second full node built by Valar Group and Tachyon. Starting with 1.0.0 (July 2026), it supports NU7 as of 1.7.0.
Wallet RPC
Zallet · Zecd
They replace zcashd’s built-in wallet. Zallet is in beta, and its first version does not support Sprout. Zecd has been audited by Zellic and supports NU7.
Fees
ZIP 317 revision
The minimum fee per logical action was lowered to 1,000 zatoshi and the weight cap was raised from 4 to 10. It has been in place since Zebra 6.4.0, and a ZIP defining a standard fee estimation method is also in preparation.
The date and scope of NU8 have not been set. These are the candidates mentioned so far.
New transaction format (ZIP 248)
A proposal for an extensible transaction format has been merged for NU8. Features such as the burn field assume this format.
NSM burn field (ZIP 233)
A field that lets anyone send ZEC to the reserve. It was left out of NU7 because it needs a transaction format change.
The next structure for development funding
What to do with the ZCG (community grants) share after NU8, including a proposal to redirect it to NSM, is being discussed on the forum. Development funding →
Memos and fees
Memo bundles (ZIP 231) and explicit fees (ZIP 2002) are in Draft status.
New discussions are taken up first on the biweekly Arborist calls.