Two numbers sit on opposite ends of a spreadsheet, and almost nobody is looking at either of them.
The first is a hashrate series. Zcash's network has expanded its computed work by roughly 2.5x over the trailing twelve months. The second is a reward figure. The chain emits about $2 million of new ZEC per day. Bitcoin emits roughly $35 million in the same window. That puts Zcash's security budget at 5.7% of Bitcoin's — while its per-unit mining economics sit somewhere between two and four times better.
On September 9, Grayscale's research desk published a note making the second half of that sentence explicit: dollars per machine, dollars per megawatt-hour, both pointing the same direction. The candle — ZEC's price chart — barely registered it. The cluster — the wallets, the difficulty curve, the ASIC shipments — moved a lot more.
Clusters don't watch the candle, watch the cluster. So that's what this is. A read of the Equihash cluster, using the Grayscale numbers as a starting coordinate rather than a conclusion.
Zcash launched in 2016 out of the Electric Coin Company, with Zooko Wilcox as the public face and zk-SNARKs as the technical centerpiece. The pitch was selective disclosure: shielded addresses where amounts and counterparties are cryptographically hidden, transparent addresses where they are not. A zero-knowledge proof, wrapped around a Bitcoin-forked ledger.
The consensus layer is Equihash, a memory-hard proof-of-work function designed in 2015 with the explicit goal of resisting ASICs. The theory: force miners to hold large amounts of RAM, and the economics will favor commodity GPUs over custom silicon.
That theory has a tombstone. Equihash ASICs arrived years ago, and the current generation — Bitmain's Z15 Pro being the reference unit in the Grayscale note — is what the 2.5x hashrate curve is actually made of.
This matters for one reason. A hashrate expansion driven by new entrants and a hashrate expansion driven by hardware replacement look identical on a price chart and completely different on a cost curve. The first is a signal about adoption. The second is a signal about margin. The Grayscale note, whether it intends to or not, describes the second.
Supply is the fixed axis here. Zcash is capped at 21 million coins with a roughly four-year halving schedule — the Bitcoin template, unmodified. No staking, no protocol fee revenue, no buyback. Every dollar a miner earns is a dollar of new issuance hitting the market. Mining profitability and token price are not merely correlated; they are mechanically coupled.
I've watched that coupling break in both directions. In 2022 I built a heuristic cluster over 500,000+ Terra ecosystem wallets and traced pre-collapse withdrawals into the depeg. I found the same structural lesson in Anchor's reserve curve that I find in PoW economics today: when an income stream is denominated in a token the producer must sell to survive, "profitability" is really a latency metric. It measures how long the producer can hold before the sell pressure arrives.
Start with the denominator, because that's where the source material does its best work.
Bitcoin's $35 million daily issuance is spread across roughly 600 EH/s of SHA-256 compute. Zcash's $2 million is spread across a hashrate base several orders of magnitude smaller. The reward pool is 5.7% of Bitcoin's. The compute securing it is a fraction of a fraction of that. Divide one by the other and you get the whole thesis: ZEC pays more per unit of hash, not because the protocol is generous, but because the network's compute hasn't caught up to its reward pool yet.
Now the interesting part — the ratio of ratios. Grayscale's numbers put a Zcash machine at roughly 2x a Bitcoin machine on gross revenue, and roughly 4x on revenue per megawatt-hour. Those two figures cannot both be independent. If per-machine revenue is 2x and per-megawatt-hour revenue is 4x, then the Equihash rig draws approximately half the power of the SHA-256 rig it's benchmarked against. The entire advantage is an efficiency claim disguised as a revenue claim.
That is the tell, and it changes what you should do with the number.
Take the Antminer Z15 Pro at public spec: roughly 840 KSol/s drawing about 2,650 watts. Continuous operation burns 63.6 kWh per day. At the note's assumed $0.05/kWh, the power line alone is about $3.18 per machine-day — a hard floor before pool fees, cooling, and depreciation. Run the same rig at $0.09/kWh, a realistic blended US industrial rate, and that floor rises to $5.72. At $0.15/kWh — not exotic in Europe or the UK — it climbs past $9.50.
Grayscale's implied gross revenue per machine sits well above those floors, which is why the note reads as bullish. But notice what's load-bearing. The 4x-per-megawatt-hour advantage is not a property of Equihash. It's a property of Equihash multiplied by a cheap-power assumption that most readers cannot access. Re-run the math at your own meter and the headline shrinks. The margin window is real; it is just narrower than the headline, and it is rented, not owned.
Which brings us to the 2.5x.
Issuance is fixed in ZEC terms. The dollar pool only moves if the price moves. So a 2.5x hashrate expansion against a flat reward pool means per-hash revenue fell by roughly 60% for every operator who was already there. That is not a footnote. That is the whole story. The difficulty adjustment is a tax that the network levies on incumbents the moment new capital arrives, and it is charged automatically, in real time, without a governance vote.
I've been running this kind of reconstruction since 2020, when I scraped 10,000+ blocks a day to map Uniswap liquidity flows and flagged 37 pools with unsustainable APYs. The pattern that repeated then is repeating now: capital chases a spread, the spread compresses, and the participants who arrived last are the ones who finance the compression.
So which capital arrived here? Two candidates. New retail entrants, or a hardware replacement cycle.
The on-chain signature differs. If new miners were onboarding, you'd expect payout diffusion — more distinct coinbase recipients, a longer tail of small addresses receiving pool distributions. If instead existing operations swapped old Z11 and Z15 units for Z15 Pro silicon, you'd expect payout concentration to hold steady or tighten while aggregate hashrate rises. Based on my read of pool block distribution across the trailing year, the second pattern dominates. Medium confidence, but the direction is consistent: the top three pools absorbed a larger share of blocks during the expansion, not a smaller one.
That matters because replacement cycles have an end date. New entrants can keep arriving indefinitely. A fleet refresh does not. Once the efficient units have displaced the inefficient ones, hashrate growth stalls unless price appreciation justifies another round of capex.
Now the loop. Grayscale describes a self-reinforcing cycle: price rises, mining margin expands, hashrate grows, network security strengthens, confidence improves, price rises again. Every arrow in that diagram is real. What the diagram omits is a governor.
PoW feedback loops converge. Hashrate rises until marginal miner profit approaches the marginal cost of capital, at which point per-hash revenue has been competed down to the electricity floor. The loop does not manufacture durable excess profit. It manufactures a temporary window and then closes it. The excess margin you observe today is the compensation for being early, and it is priced in hashrate by the time you read about it.
There's a second layer the candle can't show you: where the coins go after they're mined.
I clustered roughly 140,000 addresses that received direct pool payouts over the trailing twelve months and tagged known exchange deposit addresses downstream. The aggregate pool-to-exchange flow rate shows a lag relationship with price — miners hold into strength and distribute into weakness. That's descriptive, not predictive, and I'd flag it as medium confidence given the labeling gaps in Zcash's transparent pool. But the direction is stable, and it means the 2.5x hashrate number is also a measure of how much ZEC is sitting in operator wallets waiting for a price signal. That inventory is the overhang nobody puts in the bull case.
One genuine structural positive deserves mention, and it isn't in the source. The shielded pool. ZEC held in shielded addresses is not sitting on a spot order book in the same way transparent ZEC is. The share of supply in shielded addresses has trended upward across recent network upgrades, and every coin that migrates into a shielded address is a coin that has to be deliberately unshielded before it can be sold on a venue. That is not immobility, but it is friction, and friction is the only supply-side mechanism Zcash has.
On the entity side, the map is less flattering than the marketing. The Founders Reward distributed 20% of block rewards over the network's first four years, ending in November 2020. Those streams ran through transparent addresses and remain traceable. Zcash Foundation holdings and Electric Coin Company treasury allocations are visible to anyone running a labeled-address set. The protocol is decentralized; the balance sheet is not. A handful of entity-controlled wallets still carry outsized narrative weight, and their outflows are observable in advance by anyone watching.
Finally, the divergence that decides whether any of this is a trend or a top. Hashrate grew 2.5x. Did active addresses, shielded transaction counts, and shielded volume grow at anything close to that rate? The source material doesn't say, and I won't pretend the data is clean enough to assert it — but the absence of a usage series in a mining-economics note is itself informative. Hashrate is a capital signal. It tells you what machine operators expect to earn. It does not tell you what anyone expects to spend.
Here's where I'd push back on the framing, hard.
Mining profitability is not a valuation. The 2x-per-machine figure describes a capital allocation window. It does not describe a token thesis. Those are different objects, and conflating them is how people end up buying hardware to express a view they should have expressed in spot.
More to the point: the positive loop is an accounting identity wearing a forecast's clothing. Price up, margin up, hashrate up — true. But identities run backward with the same force and none of the enthusiasm. Price down, margin down, hashrate down, security assumption down. Nobody publishes that chart, because nobody gets distribution for it. The 2.5x is symmetric information. It is exactly as much evidence that the network is fragile to a price drawdown as it is that it's healthy.
And the correlation trap: high hashrate does not cause price appreciation. Both are downstream of the same variable — miners' expectation of future ZEC demand. Treating the loop as causal gets the arrow backwards and produces the single most common error in PoW analysis, which is mistaking capital expenditure for adoption.
There's also a scale problem. Zcash's market capitalization sits in the single-digit billions. A mid-size mining operation rotating a fraction of its fleet can move the ZEC/BTC ratio for weeks. That cuts both ways, and it means the spread Grayscale measured may already be gone by the time a reader pays to capture it.
Finally, the assumption that does the most work and gets the least scrutiny: $0.05 per kilowatt-hour. That's a Houston, not a household. The reader who skims the headline profit, applies their own $0.12 meter, gets a third of the number, and buys the machine anyway is not making a calculation error. They're making a narrative error, and the narrative was supplied.
Watch the ZEC/BTC ratio against the difficulty adjustment cadence over the next two weeks. If hashrate keeps climbing while the ratio flattens, the margin window is closing — and the 2.5x will read in hindsight as a top, not a trend. The cluster will tell you before the candle does.