I keep coming back to the same question whenever a privacy coin schedules a hard upgrade: is this a real network change, or just another date that slips once testers hit the first messy edge case? Zcash now has an answer that looks unusually concrete. Engineering groups across the ecosystem have lined up behind a November 5 mainnet target for NU7, with a testnet flip planned for October 6 and a last call on October 20. That is a tight calendar. It is also the kind of calendar that only works if the code is already close to finished.
What The NU7 Plan Actually Puts On The Table
The package is not a vague “performance update.” Teams agreed on a defined scope: faster blocks, a shutdown of version 4 transactions, and a Network Sustainability Mechanism that keeps the familiar halving path while parking fee-derived coins for later. Wallets, if current guidance holds, should not need a dramatic rewrite just because NU7 lands. Full nodes, indexers, and explorers are another story. Consensus changes always land hardest on the plumbing.
In my experience, that split matters more than the headline date. Users notice a quicker first confirmation. Operators notice three times as many blocks, new action caps, and a Sprout door that no longer opens. The interesting part is how those pieces were chosen. Coinholders voted in large numbers. Then the engineering orgs treated the results as a mandate rather than a suggestion.
A Timeline That Leaves Little Room For Drift
The working schedule is blunt. Features meant for NU7 need to be in place by September 30. Testnet activation follows on October 6. Developers then watch the network for about two weeks. On October 20 they decide the mainnet activation height. November 5 proceeds only if that review looks clean.
That last sentence is the one people skip. November 5 is a target, not a carved inscription. I like that honesty. Crypto calendars often pretend certainty. This one admits the testnet still has veto power.
- September 30: feature cutoff for code that wants a seat in NU7
- October 6: testnet activation
- October 20: mainnet height decision after live review
- November 5: current mainnet target if the review holds
Perhaps the most interesting aspect is the claimed unanimity. Multiple engineering groups signed off on both contents and timing after the latest sentiment collection. Unanimous agreement in open-source land is rare enough that I raise an eyebrow, then read the fine print. The fine print still says the deployment specification can move. Drafts remain drafts until someone publishes the final package.
Why 25-Second Blocks Are The Crowd-Pleaser
Zcash currently aims for a 75-second block interval. The draft behind faster blocks wants 25 seconds. Same issuance over a day. Smaller subsidy per block. More frequent beats on the clock. That is the whole trick, and it is a neat one if the network can absorb the extra chatter.
Why bother? Payments. Exchange deposits. Bridges that sit around waiting for a confirmation that currently feels sluggish. A first confirmation after 25 seconds, on average, changes the feel of the chain even if finality philosophy does not magically improve. People do not wait for “probabilistic settlement theory.” They wait for a number to turn green.
Faster first confirmations matter most at the edges: deposits, withdrawals, and anything that treats one block as a social signal rather than a mathematical finish line.
Throughput estimates in the draft are not shy. For simple two-action Orchard traffic, modeled capacity climbs from roughly 2.9 transactions per second toward 6.6. That is not a miracle. It is arithmetic plus tighter action limits. The same draft caps a block at 330 actions across pools, with no more than 330 Orchard actions, 300 combined Sapling inputs and outputs, and 25 Sprout JoinSplits. Limits like that are the adult supervision in the room. Faster blocks without caps would just invite larger, uglier blocks.
There is a bandwidth story too. Under a specified denial-of-service model, maximum shielded-wallet sync demand falls from about 271 MB per day to 169 MB. Wallets still pick up extra compact-block header weight, on the order of 200 KB more per day. Full nodes process three times as many blocks over the same stretch of calendar time. Nothing is free. The question is whether the fee for speed is paid in stale blocks or in operator patience.
Stale Blocks, Forks, And The Cost Of Hurrying
Theoretical stale-block risk around 3.26% at 25-second spacing is the paper number. A geographically spread devnet with 99 nodes and 2 MB blocks produced a 4.86% stale rate and a 0.37% fork rate. Those are not identical worlds. They are close enough to make me sit up. A few extra stale blocks are a tax. A fork rate that stays tiny is the number I care about more.
I’ve found that networks rarely fail on the average case. They fail when latency clusters, when a region lags, when a large block and a small block race at the worst possible second. Testing with distributed nodes is the right instinct. Two weeks on testnet after October 6 is still a short exam. Useful, yes. Complete, no. That is why the October 20 review exists.
| Setting | Current Target | NU7 Draft |
| Block interval | 75 seconds | 25 seconds |
| Daily issuance shape | Halving path | Same path, smaller per-block subsidy |
| Modeled Orchard TPS (2-action) | About 2.9 | About 6.6 |
| Action cap per block | Existing rules | 330 total, with pool-specific ceilings |
| Header overhead for wallets | Baseline | Roughly +200 KB per day |
Is 25 seconds the perfect number? Maybe not. It is a number the community was willing to defend in a vote, and it is a number engineers can test without turning the chain into a chatty toy. That combination is rarer than maximalists admit.
Halvings Stay. Reissuance Waits Until 2031.
One of the louder arguments around this cycle was issuance design. Smooth the curve, or keep the familiar cliffs? Coinholders were not split in a polite 55-45 way. Participating balances on the halving question landed near 98.9% in favor of keeping halvings. That is not a vibe. That is a slammed door.
The Network Sustainability Mechanism is the other half of the money story. The selected reading of the polls parks reissuance of removed coins until February 2031. Support for that later date was also lopsided, around 96.6% among participating balances on that question. A smaller slice wanted coins back as soon as possible. An even smaller slice liked February 2027. The conservative date won, and the engineering groups said they would treat it as the mandate for now, with room for another vote later.
Here is where people mix documents. An older smoothing draft still talks about an exponential-decay subsidy. That is not the configuration teams say they are shipping after the September vote. NU7 talk now points to an alternative NSM path: keep halvings, collect a large share of fees out of circulation, and only later feed those coins back through subsidies. Confusing the two papers is how commentary goes sideways.
At least 60% of transaction fees would leave circulation under the NSM design, with the rest available to miners. Removed coins are meant to return later through subsidies, not vanish forever.
Think about the time gap. If fee-derived coins start stacking in 2026 and only re-enter in 2031, that is years of intentional tightness. Some will call it discipline. Some will call it a long IOU. Both can be true. Markets price stories faster than protocols settle debates.
The Vote Was Large Enough To Matter
Governance theater is easy when turnout is thin. This round was not thin. Nearly 2.4 million ZEC took part, about two-thirds of the eligible Ironwood balances at the snapshot. That does not make every preference sacred. It does make it hard to pretend the result was a handful of loud accounts.
On the issuance question, roughly 2,375,932 ZEC backed existing halvings against about 22,385 ZEC for a smooth curve. On the reissuance clock, about 2,319,644 ZEC favored February 2031, while about 70,240 wanted the earliest possible return and about 6,283 preferred February 2027. You can dislike the options. You cannot honestly call the tally a coin toss.
I’ve sat through enough protocol arguments to know that “the community wants X” is often a slogan. This time the slogan has a denominator. That changes how I read the November target. The date is engineering. The contents are political in the narrow, coin-weighted sense.
Version 4 Shutdown And The Quiet End Of Sprout Spending
NU7 is expected to disable version 4 transactions at activation. That sounds like housekeeping until you remember what v4 still allows. The newer v5 format does not speak Sprout. Turn off v4, and remaining Sprout funds cannot move under current rules.
This is not a burn in the theatrical sense. Balances are not declared unissued. The spec leaves a philosophical door open for some future recovery path. Nobody is promising that path. If you still have value sitting in the original shielded pool, hope is not a migration plan.
The pool is small. Governance material put it under 23,000 ZEC and under 0.1% of transaction volume when the questions went out. Coinholders still voted for immediate deactivation at NU7. That tracks the long march from Sprout to Sapling to Orchard and into the Ironwood era that arrived with NU6.3 earlier in 2026. Legacy code is romantic until it is an attack surface.
- Identify any remaining Sprout balances before activation, not after.
- Move value into supported pools while v4 still works.
- Treat “maybe a recovery tool later” as a rumor, not a product.
- Assume explorers and wallets will show a dead end, not a refund button.
Is it harsh? A little. Is it surprising after years of disuse? Not really. Privacy systems age in public. Old circuits become museum pieces with keys still attached. Closing the exhibit is overdue if the collection is almost empty.
Who Has To Change Software, And Who Can Sit Tight
Guidance from the rollout discussion is unusually kind to ordinary wallets. No new transaction format is part of the announced scope. If you already speak the current generation of shielded transfers, NU7 itself may look like a consensus patch rather than a product rewrite. That is the optimistic read.
Operators should not share that optimism. Faster blocks change indexing assumptions. Action limits change block validity. v4 deactivation changes what a “historical spend” even means. Explorers that still render Sprout paths as live will confuse users on day one. Node software that lags the activation height will fork itself into a museum.
Zebra’s recent production line already lives on NU6.3. Node hardening through 2026 included a wider local rollback window, from 99 blocks to 1,000, as a defense against long consensus splits. That is the unglamorous work that makes a 25-second world less terrifying. I would rather see boring node resilience than another marketing diagram about “instant privacy.”
NU7 operator checklist - Confirm node release tracks the final deployment spec - Revisit indexer assumptions for 25-second cadence - Update explorers for v4 / Sprout spend halt - Watch stale-block and fork metrics on testnet - Do not assume wallet UX is the only surface that moves
The Market Rally Is Real. It Is Not A Proof.
ZEC has been loud. Prices near the mid-1,400s on September 18, a jump of roughly 9.6% over 24 hours, volume above 2.6 billion dollars. A day earlier the token had already ripped more than 20% toward the 1,300s. Earlier in September it crossed 1,000 after a U.S.-listed investment product conversion drew fresh attention and a starting bag in the low hundreds of millions of dollars in assets.
None of that proves NU7 is priced to perfection. Upgrade optimism is one ingredient. Institutional wrappers are another. Momentum traders do not wait for October 20. They trade the calendar. That can help a narrative. It can also punish a slip with theatrical precision.
In my view, the healthier way to read the tape is as a reminder that attention arrived before the testnet did. Attention is fuel. Testnets are brakes. You want both.
What “Draft” Still Means In Practice
The ZIP shelf still labels several NU7 ideas as candidates. The deployment document is supposed to freeze the package. Until that freeze is public and boring, the October 20 meeting is not ceremonial. Features that miss September 30 can be kicked. Specifications that misbehave on testnet can be trimmed. That is the process coinholders were told to expect.
I prefer that messiness to a fake sense of completion. A privacy network that ships a half-tested interval change is not being brave. It is being casual with other people’s confirmation risk. Two weeks of testnet will not catch every long-tail bug. It can catch the obvious ones, and the obvious ones are how upgrades embarrass themselves.
A target date without a review date is marketing. A review date without the courage to delay is also marketing.
How Faster Blocks Change Everyday Use
Imagine sending value to an exchange that wants one confirmation. Today that wait clusters around a minute and a quarter. After NU7, the same policy clusters around half a minute. Merchants who treat one block as “good enough for small tickets” will feel the difference first. Cross-chain operators who batch on confirmation count will feel it next. People who wait for many blocks for large transfers will notice less, because six fast blocks are still six blocks.
There is a psychological trick here. Humans hate idle progress bars. A chain that ticks three times as often looks alive. That can improve perceived reliability even when deep reorganization risk is a separate conversation. Perceived reliability still moves deposits. I will not pretend otherwise.
The action caps are the quiet user-protection story. A world with more blocks and no ceilings would invite pathological transactions that punish every compact-client sync. The draft tries to raise honest throughput while cutting a modeled worst-case bandwidth number. That trade is the grown-up version of “make it faster.”
Privacy Design Does Not Pause For A Clock Change
Shorter intervals do not automatically weaken shielding. They also do not automatically strengthen it. Privacy lives in circuit choice, note management, and whether users actually shield. A faster block time can help adoption if payments stop feeling sticky. It can hurt operational privacy if wallets leak metadata while racing to keep up. Implementation quality decides which way that cuts.
The migration away from Sprout is part of that hygiene. Old pools with tiny volume and ancient assumptions are not a feature. They are a footnote that still needs a key. Turning off the spend path is cleaner than leaving a dusty side door unlocked “just in case.”
Ironwood already shifted the architecture earlier in 2026. NU7 is less of a brand-new house and more of a renovation: faster stairs, a locked basement, and a maintenance fund with a delayed payout. Renovations are less exciting than groundbreakings. They are how houses stay standing.
Risks That Deserve A Straight Sentence
Stale blocks can rise. Mining variance can look noisier at a shorter interval. Indexers can fall behind if they were written for a leisurely 75-second heartbeat. Users with forgotten Sprout funds can discover a dead end. Markets can treat any October delay as a broken promise even if delay is the responsible move.
There is also documentation risk. Multiple drafts, an older smoothing design, and a post-vote alternative can live side by side on the same repository shelf. Casual readers will flatten them into one story. That is how rumors about “issuance changing next month” get born. The selected path keeps halvings. Reissuance, if the NSM rules survive review, waits until 2031.
- Consensus bugs show up as splits, not as blog posts
- Wallet silence is not the same as infrastructure readiness
- A high vote share does not freeze cryptography
- Price strength can vanish if testnet metrics disappoint
None of those risks make the upgrade foolish. They make the October review non-negotiable. I would rather read a dull delay notice than a dramatic incident report.
How I Would Watch The Next Six Weeks
First, treat September 30 as a scope filter. If a shiny extra misses the cutoff, good. Scope discipline is how upgrades ship. Second, watch October 6 like an operator, not a spectator. Testnet is not a parade. It is a lab. Third, wait for the October 20 height decision before writing November 5 into muscle memory.
On the product side, I would audit addresses the way a skeptic audits a closet. Anything still living in Sprout needs a move while movement is legal. Custodians should publish readiness notes in plain language. Explorers should preview how a disabled v4 path will render. Silence from infrastructure is a smell.
On the money side, I would separate the NSM story from the price chart. A 2031 reissuance clock is a long-duration policy. It should not be used as a one-week trading thesis. Halvings remaining in place is the nearer issuance fact. Fees leaving circulation is the nearer supply fact, assuming the rule survives review.
Why This Upgrade Feels Different From A Generic Hard Fork
A lot of protocol upgrades are feature dumpsters. New opcode here, new format there, a mascot in the release notes. NU7 is narrower. It tries to make the chain feel quicker without inventing a fresh transaction dialect. It tries to close an old pool without pretending the coins were never there. It tries to install a fee sink without ripping out the halving folklore people already understand.
That combination is oddly conservative for a project that still carries a radical privacy brief. I mean that as a compliment. Privacy coins do not need more mythology. They need confirmations that arrive, nodes that agree, and rules that match the last vote.
Will November 5 hold? Ask again on October 20. Until then, the story is not “Zcash is done.” The story is “Zcash picked a package and agreed to test it in public.” That is a better story than a date shouted into a void.
A Practical Close For Holders And Builders
If you hold ZEC and use modern shielded paths, your homework is mostly awareness: dates, node versions, and the Sprout footnote. If you build on the chain, your homework is heavier. Re-profile sync. Re-check confirmation UX. Rehearse the v4 cutoff. If you trade the token, remember that calendars are not cash flows. A testnet surprise can reprice a narrative in an afternoon.
The next checkpoint is already on the wall. Code complete enough by September 30. Testnet live on October 6. Adults in the room on October 20. November 5 only if those adults like what they see. That is the whole plot, and it is enough plot for now.
I keep my own bias simple. Faster blocks are worth the operational tax if stale rates stay honest and wallets do not drown in headers. Keeping halvings matches what holders actually voted. Parking NSM reissuance until 2031 is a long bet on discipline. Shutting v4 is housekeeping with teeth. Put together, NU7 is less a revolution than a tightening of bolts. Sometimes that is exactly what a live monetary network needs.
