Zcash has activated its NU7 network upgrade on public testnet, ahead of a mainnet window still aimed at November 5. Developers are due to sit with the test results on October 20 and decide whether the same rules should cross over, and at which height. Nothing on the production chain has changed yet. That sentence matters more than the headline, because testnet excitement has a habit of leaking into price chatter before wallets, exchanges, and half-forgotten shielded notes have caught up.
A Faster Clock, With the Issuance Math Left Intact
NU7 cuts the target block spacing from 75 seconds to 25 seconds. In plain terms, the network tries to produce about three blocks in the time it used to aim for one. The average wait for a first confirmation drops with it, from a minute and a quarter to something closer to half a minute. I have always thought Zcash felt slightly out of step with how people actually move money, not because privacy is slow, but because 75 seconds is a long stare at a pending spinner when you are only moving a modest amount.
Shorter blocks are not free. They change orphan risk, they change how much work a single block is allowed to contain, and they change the rhythm exchanges use when they count confirmations. The engineering answer, at least on paper, is a set of new limits on shielded actions so a block cannot be stuffed with so much proving work that wallets choke. More frequent blocks, tighter caps on what each one may carry. That trade is the part I would watch on testnet before anyone celebrates throughput.
There is a second adjustment that keeps the upgrade from becoming an accidental inflation event. Reducing the interval does not triple the amount of ZEC issued over the same calendar period. Issuance is rescaled alongside the shorter schedule, so miners are not handed three times the daily subsidy just because the chain is ticking faster. If you have seen the “triple the blocks, triple the coins” take already circulating in chats, file it under unfinished arithmetic.
What Actually Activated, and When
The activation block was mined on October 4, earlier than the rough October 6 estimate developers had been using. The project later confirmed that NU7 was live on testnet. Operators running the release candidate were asked to exercise the upgrade under real conditions. Mainnet operators were told, correctly, that they did not need to change anything yet, because no production activation height had been set.
That gap between testnet and mainnet is the whole point of the next two weeks. A consensus change can look clean in a lab and still misbehave once public traffic, odd wallets, and delayed peers show up. October 20 is the review date. November 5 remains the target, not a promise. I would treat any countdown clock you see on social feeds as marketing until an activation height is actually published.
A testnet block is a rehearsal, not a coronation. The useful question is what broke quietly, not what trended.
The preparation had been public for weeks. Engineering teams settled on the proposed NU7 package in September and pointed at November 5 only after the testnet phase. Node software released at the start of October added testnet support. Earlier September builds had already been carrying related groundwork. None of that, by itself, forces mainnet. It does mean the client side is no longer guessing.
Why 25 Seconds Won the Room
Before the schedule was locked, a private governance poll covering nearly 2.4 million ZEC came back heavily in favor of the shorter interval. Roughly 99.9 percent of participating coins backed a 25 second target. Participation was about 66 percent of the eligible balances tied to the current shielded pool used for the vote. That is not the entire supply, and private polls are not chain law, but it is hard to call the direction ambiguous.
Two related questions landed in the same exercise. Around 98.9 percent supported keeping the existing halving schedule. Approximately 96.6 percent backed February 2031 as the start of reissuance under the new sustainability mechanism. Holders, at least the ones who showed up, wanted faster blocks without rewriting the long-run issuance curve they already understood.
I find that combination more interesting than the block-time headline. Privacy coins live or die on whether long-term holders trust the supply story. Speeding up confirmations is a user-experience win. Touching issuance is a credibility risk. NU7 tries to do the first while leaving the second recognizably intact, then adds a fee recycle on the side. Whether that recycle feels like stewardship or like a quiet tax will depend on how clearly wallets explain it.
The Network Sustainability Mechanism, Without the Brochure
The other pillar of NU7 is the Network Sustainability Mechanism, usually shortened to NSM. Under it, part of each transaction fee is pulled out of immediate circulation and placed into a reserve that can fund later block rewards. The plain description used by the implementation teams is that the system recycles a slice of fees into future issuance.
Think of it as a split, not a burn in the casual meme sense, and not a miner tip that vanishes into the next block either. Some of the fee still compensates the producer of the current block. Some of it is parked so that, years out, when the subsidy has stepped down again, there is a pool to draw from. February 2031 is the date the poll attached to the start of that reissuance path.
Why bother? Zcash, like most proof-of-work chains with a declining subsidy, eventually has to answer a dull question: who pays the people securing the ledger once new coins thin out? Bitcoin’s answer is fees, full stop. A privacy chain has a harder version of the same problem, because shielded usage can be lumpy, and because users already pay a cognitive tax just to use the private path. NSM is an attempt to smooth that future instead of hoping fee markets mature on schedule.
Perhaps the most interesting aspect is what it does to miner expectations in the near term. Today’s block is slightly less fee-rich than it would have been under a “miner takes all” rule. Tomorrow’s subsidy is slightly better supported if usage actually generates fees worth recycling. If usage stays thin, the reserve stays thin. Mechanism design cannot invent demand.
- A portion of each fee leaves immediate circulation instead of paying only the current block producer.
- That portion sits in a reserve earmarked for later block rewards.
- The halving cadence holders already know is meant to stay in place.
- Reissuance from the reserve is aimed at February 2031, not at the November activation.
- Daily new supply is not tripled just because blocks arrive three times as often.
If you are modeling miner revenue for the next quarter, the subsidy rescale matters more than NSM. If you are modeling 2032, NSM is the line item. Mixing those horizons is how people talk past each other on upgrade threads.
Shielded Action Limits, and the Spam Problem Nobody Posts About
Faster blocks sound like pure upside until you remember what a shielded transaction actually costs a node. Proofs are heavy. A block that arrives every 25 seconds cannot be allowed to carry an unbounded pile of shielded actions, or wallets and modest nodes start falling behind the tip. NU7 adds limits designed to cap that work per block.
The stated goal is to protect wallets from spam while block production becomes more frequent. I buy the motive. I am less sure the first set of caps will feel right to power users who batch a lot of private notes. Testnet is exactly where that friction should show up: a wallet that used to clear a large shielded send inside one block may now need several. That is not a failure. It is the design. It will still annoy people who were not in the room when the cap was chosen.
There is a user-facing consequence worth saying out loud. Confirmation time falls. The amount of private work you can stuff into a single confirmation may also fall. Net latency for a simple payment should improve. Net latency for a fat, many-output shielded transaction might not, if it has to span blocks. Anyone writing “three times faster” without that caveat is selling a slogan.
Version 4 Transactions, and the Sprout Trap
This is the section I would pin to the top of any holder note. Once NU7 reaches mainnet, version 4 transactions are set to be disabled. Spending from the legacy Sprout shielded pool depends on those version 4 transactions. Coins left in Sprout would become unspendable after activation unless the holder moves them first.
No such mainnet change has happened. The rule is live only on the public test network. Still, “later” is a poor plan if your notes sit in a pool you have not opened since the early shielded era. Sprout was the first private pool. Most active use migrated years ago to newer circuits. The long tail is exactly who gets hurt by a hard cutoff: old paper wallets, dusty exchange withdrawals, funds someone meant to consolidate and never did.
I have found that these migrations always look obvious in hindsight and invisible in the month before. The people who need the warning are not refreshing protocol blogs. If you custody ZEC for anyone else, the responsible move is to identify Sprout balances now, while the mainnet height is still unset, and to build a move-out window that does not depend on the final week.
Unspendable is not the same as seized. It is worse in one narrow way: there is no counterparty to argue with, only a rule your client will refuse to build a transaction for.
A practical way to think about legacy pool cutoffs
Exchanges and custodians have their own version of this problem. Customer balances may be pooled, so the user-facing label “ZEC” hides which circuit the underlying notes use. A venue that still has Sprout exposure needs a migration plan before any mainnet height is frozen, not after. The testnet activation is the alarm, not the lock.
How This Sits Next to Ironwood
NU7 is not arriving in a calm decade. It follows Ironwood, which activated on July 28 at block 3,428,143, only a few months ago. Ironwood existed because of a flaw in the Orchard shielded pool, a flaw that could theoretically have allowed undetectable counterfeit coins to be created. Researchers had not found evidence the bug was exploited. The privacy model also meant developers could not conclusively prove that unauthorized coins had never been created inside the affected pool.
That is the uncomfortable honesty at the center of shielded supply. You can audit what enters and leaves a pool at the boundary. You cannot open every note and count them the way a transparent ledger can. Ironwood answered the uncertainty by standing up a separately tracked shielded pool, with accounting safeguards on how much value could leave the older one. Orchard stayed open for withdrawals. New shielded funds moved into the new pool.
The node implementation of that pool reused the existing action structure and proof system, while keeping separate records for the note commitment tree, the nullifier set, the chain value pool, and chain history. Emergency measures in June had already disabled affected transactions before corrected software shipped. Ironwood was the longer fix.
Why retell that story inside an NU7 piece? Because holder trust is path-dependent. A network that just had to split a pool over a theoretical inflation bug is now asking people to accept a fee reserve and a legacy-pool cutoff. Both can be justified. Neither should be waved through on vibes. In my experience, the chains that keep sophisticated users are the ones that narrate supply changes in boring detail, repeatedly, until the boring detail is what people remember.
Node Software, a Crash Bug, and Who Is Already Moving
Infrastructure has been noisier than the consensus proposal. A high-severity issue in two recent node releases could let a peer remotely crash a node. The fix shipped on September 25. The project said the bug hit availability, not consensus and not funds. Availability bugs still matter in the week you are asking operators to trial a new upgrade. A node that falls over cannot tell you whether the new rules are healthy.
There is also a client-diversity subplot. One large exchange moved its Zcash infrastructure onto an alternative stack ahead of NU7 and reported reaching the chain tip in a little over six hours, against close to a day on the more common implementation, while deposits and withdrawals kept running. I do not read that as a verdict on either codebase. I read it as a reminder that sync time and operational headroom will shape who is ready on day one of mainnet, if mainnet happens on the current target.
Release candidates are exactly that. Running one on testnet is the job. Running one on mainnet before an activation height exists is optional enthusiasm. The October 2 candidate was aimed at public testnet. Production operators can wait for the October 20 read, then for a tagged release that names a height. Patience here is not laziness.
What Changes for a Normal Payment
Strip the governance language off and a retail send looks like this. You broadcast. You wait. Under the old target you were staring at roughly 75 seconds for the first inclusion attempt to land in expectation, longer if the mempool was awkward. Under NU7 the target wait is about 25 seconds. Exchanges that require several confirmations would see their internal timers shrink only if they keep the same confirmation count. Some will raise the count to preserve the same wall-clock finality. That choice is theirs, and it will not be uniform.
Fees get a new split. The sender still pays. The miner does not keep every satoshi of that payment. A slice is reserved. Wallets that only show “network fee” without explaining the split are going to generate support tickets. This is a product problem as much as a protocol problem, and product problems are where upgrades go to feel broken even when consensus is fine.
Shielded sends pick up the action caps. A simple private payment should feel snappier. A complex one may clear across more than one block. Transparent sends are less affected by proving limits, though they still live on the faster clock and the new fee rule. If your use of ZEC is mostly transparent, NU7 is a latency and fee-policy change. If your use is shielded, it is those things plus a capacity shape you should test.
| Piece of NU7 | What it changes | Who feels it first |
| 25 second target | Shorter average wait for first confirmation | Wallets, exchanges, payment flows |
| Issuance rescale | Daily new ZEC stays on the old economic path | Miners, supply models |
| Action limits | Caps shielded work inside one block | Heavy private users, wallet devs |
| Fee reserve | Part of each fee funds later rewards | Miners now, subsidy path after 2031 |
| Version 4 cutoff | Sprout spends stop after mainnet activation | Legacy shielded holders |
That table is the upgrade in one glance. Everything else is timing, software, and whether October 20 still likes what testnet is showing.
Miners, Orphans, and the Quiet Cost of a Shorter Interval
Proof-of-work does not become easier because you asked blocks to arrive sooner. Difficulty retargets. The expected work per unit time stays in the same family, spread across more block events. What changes is variance and propagation. A 25 second target leaves less slack for a block to reach the whole network before someone else finds a competing one. Orphan rates can rise if propagation does not keep up.
I am not predicting a mess. Zcash blocks are not enormous, and the action caps are partly there to keep them from becoming enormous. I am saying the testnet metric that deserves a chart is stale-block rate, not social engagement. If orphans climb and stay climbed, miners will feel it as lost revenue even with the subsidy rescale designed to hold daily issuance steady. A rescale that assumes blocks mostly get included is a rescale with a footnote.
Pools will also rewrite their payout math. Shares, variance, and the fee split all move. Smaller miners who were already marginal do not get a gift from faster blocks. They get a different noise profile. Anyone telling you NU7 is automatically good for hash rate is guessing.
A Calendar You Can Actually Use
Dates are the only part of this story that will be wrong if you memorize them too confidently, so hold them loosely.
- September: teams agreed the NU7 package and a November 5 mainnet target after testnet.
- September 25: a node crash bug was patched, availability only.
- October 2: a release candidate invited testnet operators to run the upgrade.
- October 4, 18:21:45 UTC: activation at testnet block 4,465,026, earlier than the old estimate.
- October 20: scheduled review, and the moment an activation height could be chosen.
- November 5: still the aimed-for mainnet day, not yet finalized.
- February 2031: the polled start for reissuance under the fee reserve, far beyond this upgrade’s first week.
Between October 20 and any mainnet height, the practical work is dull. Update nodes when a production release says to. Confirm your wallet understands the fee split. Find Sprout notes. Ask your custodian, in writing, which pool your balance sits in. None of that requires a price view.
What I Would Watch on Testnet Before October 20
Speculation is cheap. A short checklist is not.
- Actual block interval versus the 25 second target, over days, not hours.
- Orphan and reorg frequency compared with the pre-activation baseline.
- How often shielded transactions spill across the new action caps.
- Wallet sync and proof-generation time on ordinary hardware.
- Whether the fee split is visible and correctly accounted in block explorers.
- Node crash rates on the release candidate, given the recent availability bug.
- Any surprise interaction with the post-Ironwood pool accounting.
If those lines look boring, that is the win. Upgrades that trend are often upgrades that stumbled. A quiet testnet is allowed to be the story.
Price Chatter Versus Protocol Reality
ZEC has a market, and markets front-run calendars. A testnet activation can get traded as if mainnet already shipped. It has not. The October 20 meeting can still slip the height, narrow the package, or ask for another test cycle. I have no edge on the next candle, and neither does a block explorer screenshot.
What can be said without cosplay certainty is structural. Faster confirmations are a usability positive if orphans stay tame. A legacy pool cutoff is a supply-liquidity event for the coins that move in time, and a frozen-balance event for the coins that do not. A fee reserve is a long-dated claim on future usage, not a dividend. Halving stays. Daily issuance is not tripled. Anyone’s model that ignores one of those sentences is incomplete.
Privacy coins also trade on regulatory weather, which this upgrade does not change. NU7 does not make shielded transfers more or less legal in any jurisdiction. It changes spacing, fees, action limits, and an old transaction version. Conflating a consensus tweak with a policy shift is how threads go sideways.
Wallets, Labels, and the Explanation Gap
The best protocol change in the world still dies in a settings screen. After a mainnet activation, a competent wallet needs to do a few unglamorous things. It should refuse to build a version 4 spend and tell the user why, in language that mentions Sprout by name. It should show the fee split, not a single opaque number. It should warn when a shielded send exceeds what one block will hold, and offer to split it. It should not imply that 25 seconds is a promise rather than a target.
Light clients have a harder job than full nodes here, because they already trust intermediaries for state. A faster chain means more headers, more chances for a lite client to be slightly stale, more importance on checkpoint honesty. None of that is new in kind. It is new in tempo.
If I were shipping a ZEC wallet this month, I would put the Sprout migration banner up before the marketing banner about faster blocks. Fear of looking alarmist is how legacy funds get stuck. A banner you can dismiss is cheaper than a forum post in December from someone who cannot move coins.
A Fair Criticism, Then the Reply
The skeptical read is easy to write. Zcash just restructured a shielded pool because of a serious bug. It is now stacking another consensus change on top, with a fee diversion some miners will call a skim, and a hard stop on an old pool that will strand someone. Testnet ran early, which can mean confidence or can mean the estimate was loose. Governance was a coin-weighted private poll, not a chain vote. Why rush November?
The reply is also easy, and I think it is mostly fair. The bug response is why Ironwood exists, and NU7 does not pretend to reopen that wound. The fee reserve was polled alongside the block-time change, with heavy support among coins that participated. The Sprout stop is harsh precisely because leaving a broken or obsolete spend path forever is its own risk. Early testnet activation buys review days, it does not spend them. November 5 is a target with a decision gate on October 20 in front of it.
Both paragraphs can be true. Upgrades are allowed to be justified and still deserve a suspicious reading. That is not disloyalty to a ticker. It is how you avoid learning supply rules from a price candle.
How Confirmation Policy May Shift in Practice
Merchants and exchanges rarely think in targets. They think in “how many blocks until I release the goods.” A shop that waited two confirmations under a 75 second target was waiting about two and a half minutes in expectation. The same two confirmations under a 25 second target land closer to a minute. Some risk desks will keep the count. Others will move to six confirmations to sit near the old wall-clock window. Both are coherent. Users will experience them as inconsistency, which is annoying and normal.
Reorg depth is the variable that should drive that choice, not nostalgia for the old interval. If testnet and then mainnet show that short reorgs stay rare, a lower confirmation count is rational. If short reorgs pick up because propagation lags the new tempo, raising the count is rational. Publishing the policy, either way, matters more than picking the fashionable number.
Rough expectation, not a guarantee: Old target: 75 seconds per block New target: 25 seconds per block Two confirmations, old: about 150 seconds Two confirmations, new: about 50 seconds Six confirmations, new: about 150 seconds
Those figures assume the chain is hitting its target and your transaction is not stuck behind a cap or a fee bump. Real mempools are ruder than targets. Still, the sketch is enough to explain why “faster chain” and “faster withdrawal” are not synonyms until the venue updates its policy.
Supply Narrative, Said Slowly
People buy privacy coins for different reasons, but the ones who stay tend to care about the issuance story. NU7’s story, compressed, is this. The halving schedule holders voted to keep remains the spine. Block production speeds up, and the per-block subsidy steps down so the calendar does not suddenly rain coins. Fees are partly saved rather than fully spent in the block that collected them. A future reissuance window, polled for February 2031, can draw on that savings if the design ships as described. Legacy Sprout value that is not moved becomes stuck, which is a reduction in usable supply, not a mint.
Stuck is not burned in the social-media sense. It is inaccessible under the new rules. Economically, coins nobody can spend stop competing as sell pressure. They also stop being collateral, savings, or donations. Calling that a bullish burn is a tell that the speaker has not had to explain it to someone who lost access. Say “unspendable if you do not move them” and leave the slogan merchants to their work.
Ironwood’s boundary accounting still sits underneath all of this. Value leaving the older affected pool remains constrained by the safeguards introduced in the summer. NU7 does not erase that history. A serious supply note mentions both.
What Holders Can Do This Week
You do not need to run a node to be competent about this upgrade. You do need a short list.
- Identify whether any of your ZEC is still in the Sprout pool.
- If it is, plan a move while version 4 spends are still valid on mainnet.
- If a custodian holds it, ask which pool and what their migration window is.
- Ignore activation-height rumors until a height is actually specified.
- Treat November 5 as a target behind an October 20 review.
- Update wallet software when maintainers say NU7 mainnet support is in, not before out of impatience.
- Expect fee displays to change, and read the split before you assume you were overcharged.
That is the whole retail playbook. Everything else is for operators and for people who enjoy consensus diffs, which is a respectable hobby and a poor substitute for checking your own notes.
Operators, Pools, and the Unsexy Cutover
Mining pools and public nodes carry the upgrade whether or not holders read the notes. A pool that does not upgrade by the activation height stops producing valid blocks. A public node that does not upgrade stops following the canonical chain. The testnet path is the dress rehearsal for that cutover: release candidate, activation, then a look at who fell off the tip.
The recent crash bug is a reason to prefer the patched line and to watch peer behavior, not a reason to skip the upgrade forever. Availability faults and consensus faults are different categories. Mixing them up produces either panic or complacency, and both are expensive.
Client diversity, already visible in how quickly different stacks reached the tip, is a quiet strength if more than one implementation can validate NU7. It is a quiet risk if most of the economic majority sits on a single binary. I do not have a fresh census of that majority. I do know that a network fresh off an inflation scare should want more than one set of eyes on the new rules.
A Note on Language, Because Words Move Price
“Goes live” is doing a lot of work in headlines this week. Live on testnet is real. Live on mainnet is not, yet. “25 second blocks” is a target, the way 75 seconds was a target, not a stopwatch the protocol owes you. “Fee burn” is the wrong noun for a reserve that is designed to come back as later rewards. “Governance approval” overstates a private coin-weighted poll, even a lopsided one.
I keep tripping over these because sloppy nouns are how a careful upgrade gets remembered as something else. If you write about NU7, or you repeat a thread about it, the precise version is short enough to fit in a message: testnet only, faster target, issuance rescaled, part of fees reserved, Sprout spends die on mainnet unless moved, decision gate on October 20, November 5 still aimed at and not locked.
NU7 in one line: testnet now, 25s target, subsidy rescaled, fee reserve, Sprout cutoff on mainnet, height still unset.Where This Leaves the Privacy Tradeoff
None of the timing work removes the original bargain. Shielded transfers hide amounts and participants from the public ledger, and they cost more computation than a transparent send. NU7 tries to make the wait shorter and the per-block load bounded, so that bargain is easier to live with on an ordinary afternoon. It does not make proving free. It does not make a light client omniscient. It does not answer whether a given venue will keep supporting private deposits.
The sustainability mechanism is the piece that reaches past this year’s user-experience gripe. It assumes that fees will exist in enough size, someday, to matter as a security budget. That assumption can fail. If private usage stays a niche inside an already small asset, the reserve stays symbolic. Symbolism is not fraud. It is also not a completed monetary policy. Worth saying, so nobody treats February 2031 as a scheduled paycheck.
I still think the direction is coherent. A privacy network that cannot get a first confirmation inside half a minute will keep losing casual users to transparent defaults. A privacy network that triples issuance to buy that speed would deserve the backlash. Doing the speed change and refusing the inflation shortcut is the adult version of the upgrade. The Sprout cutoff is the part that has to be executed with more warning than engineers usually enjoy giving.
Questions Worth Asking Before You Trade the Headline
Is your balance in a pool this upgrade can still spend? Has your venue said when it will enforce the new rules? Are you pricing a testnet event as if the height were already mined on mainnet? Do you know the difference between a fee that is reserved and a fee that is destroyed? If the October 20 review slips the date, does your thesis still exist?
Those are not gotchas. They are the difference between following a protocol and following a ticker. NU7 can be a genuine improvement in how Zcash feels to use and still be a bad intraday trade. It can also be a non-event on price and a real event for anyone with legacy notes. Both outcomes have happened to other upgrades, on other chains, often in the same week.
The Longer Arc, Kept Short
Zcash’s recent protocol history is a sequence of constrained choices. An Orchard flaw forced a temporary shutdown of affected spends, then a corrected release, then Ironwood’s separate pool and exit accounting. NU7 is the next scheduled step: block timing, transaction rules, shielded action limits, and the sustainability mechanism. It is not a rebrand. It is not a new asset. It is a ruleset change with a testnet birthday of October 4 and a mainnet decision still ahead of it.
If the review on October 20 likes what it sees, an activation height gets set and November 5 stays the day people circle. If the review does not, the honest outcome is a delay, and a delay would be a sign the process works. I would rather read a slipped date than a quiet reorg essay. That preference is personal. It is also how I read every chain that still does hard, scheduled upgrades instead of pretending parameter changes are weightless.
Bottom Line for Anyone Still Holding the Tab
NU7 is real on testnet, at block 4,465,026, with a 25 second target and a fee path that saves a slice for later rewards. Mainnet has not moved. Issuance is not tripling. Sprout spends are on a clock that starts only when mainnet activates, and that clock is the detail most likely to hurt a real person. October 20 is the next decision. November 5 is the hope.
Check the pool. Read the fee. Wait for a height. The rest is noise until the chain, not the timeline, says otherwise.
]]>
