On September 9, a research note from Grayscale landed with a claim that should not be possible in a market this old. A single Zcash mining rig, the note said, earns roughly twice what a comparable Bitcoin machine earns per day. Measured per megawatt-hour — the only unit of account that actually matters in mining — the gap widens to nearly four times.
I read that number three times. Then I did what I always do, which is take it apart and see whether it survives reassembly.
Because there is a second number in the same report, and it is far less flattering. Bitcoin distributes roughly $35 million every day to the machines securing it. Zcash distributes roughly $2 million. That is 5.7 percent of Bitcoin's reward pool. A network paying out one-eighteenth as much cannot, structurally, sustain a fourfold energy-efficiency advantage for long. Not because the arithmetic is wrong — the arithmetic is correct — but because the arithmetic is not the mechanism. It is the residue.
The crowd sees a moon. I see a model. And the model has a decay constant.
To understand why, you have to go back past the price chart to the algorithm itself.
The algorithm that was supposed to stay boring
Zcash launched in October 2016, built by the Electric Coin Company under Zooko Wilcox, and it arrived carrying two genuinely novel things. The first was the first production deployment of zk-SNARKs — zero-knowledge proofs used not as a laboratory curiosity but as the mechanism by which a transaction can be validated without anyone learning who sent what to whom. The second was Equihash, a proof-of-work function designed to be memory-hard rather than compute-hard.
The design intent was explicit. Bitcoin's SHA-256 is a pure arithmetic grind; it rewards whoever can pack the most hashing engines onto silicon, which is why it industrialized into ASICs within eighteen months of GPU mining beginning. Equihash instead asks miners to solve a generalized birthday problem. The 200,9 parameterization used by Zcash requires assembling a solution set across roughly 144 megabytes of memory, and the bandwidth requirement scales in a way that, in theory, keeps general-purpose hardware competitive. The promise was that mining would stay diffuse — spread across many operators running commodity machines rather than concentrated in the hands of a few industrial buyers placing nine-figure orders with a single foundry.
For about two years, that held. Then in 2018, Bitmain shipped the Antminer Z9, and the ASIC-resistance claim became a historical footnote. Today the Zcash network is dominated by the Antminer Z15 Pro and its descendants — machines purpose-built for the exact Equihash parameter set, with memory subsystems tuned to the precise bandwidth profile the algorithm demands.
I want to be precise here, because the distinction matters for everything that follows. Equihash did not fail to resist ASICs. It delayed them, and the delay shaped the network's industrial structure in ways that are still compounding. Bitcoin's hashrate is a commodity market with dozens of competing machine generations and a brutal, continuous efficiency race. Zcash's hashrate is closer to a single-vendor dependency. That is a different risk surface entirely, and the Grayscale note does not address it, because the Grayscale note is about yield.
There is a second structural artifact worth naming. Zcash has since moved through Halo 2, the upgrade that removed the trusted setup from its proving system — a genuine cryptographic achievement that eliminated one of the original protocol's most criticized trust assumptions. That work matters. It is also, from a mining-economics standpoint, irrelevant. The proving system secures the privacy of transactions. It does nothing for the security of the chain itself, which remains a pure proof-of-work ledger whose integrity is bought with hashrate and electricity. Conflating the two is a common error, and it flatters the network in ways the difficulty curve does not.
Narratives are liquid; truth is solid. The solid truth about Equihash in 2026 is this: the algorithm's memory-hardness makes the energy curve flatter than SHA-256's. Where a Bitcoin machine's efficiency improves along something adjacent to a Moore's-law slope, an Equihash machine's improvement is bounded by memory bandwidth, which improves more slowly and more expensively. A flatter efficiency curve means older Zcash machines stay profitable longer, which means hashrate responds to price signals with more inertia than Bitcoin's does.
That inertia is the actual subject of the report. Everything else is downstream.
Decomposing the 2.5x
The headline figure is a 2.5x increase in network hashrate. Anyone can read that off a block explorer. The interesting question — the one that took me the better part of a weekend to work through — is what kind of growth it is.
There are three possibilities, and they have wildly different implications for anyone trying to price ZEC.
The first is new capital entering: fresh mining operations buying fresh machines and pointing them at Zcash. This is the bullish reading, and it is the reading the report implicitly favors.
The second is substitution within existing operations: miners who already own Equihash-capable hardware redirecting it from other networks, or from idle inventory, toward Zcash because the relative yield improved. This is a reallocation, not an expansion. It costs almost nothing to do and can reverse just as cheaply.
The third is fleet replacement: operators retiring older GPUs and first-generation ASICs and replacing them with Z15 Pro-class machines. This raises hashrate without adding a single new participant to the network.
Based on the efficiency profile of the incremental hashrate — and this is inference, not disclosure, because the report publishes no machine-level breakdown — I put the majority weight on the third category, with meaningful second-category flow. The reason is timing. A 2.5x move over roughly a year is too fast for greenfield deployment, which requires capital formation, site selection, power contracts, and a nine-to-twelve-month hardware lead time. It is exactly the right speed for a fleet refresh cycle triggered by a favorable price move.
Why does this distinction matter so much? Because substitution and replacement are elastic. They arrive quickly when relative yield improves and they leave quickly when it deteriorates. Only the first category — genuine new capital, committed to a site and a power contract and a machine order — has switching costs that make it sticky.
The report frames the hashrate increase as a signal of miner confidence. I would frame it as a signal of miner opportunism. Those are not the same thing, and the difference between them is the difference between a floor and a ceiling.
There is one more mechanical detail worth surfacing, because it explains part of why the growth looks so violent. Equihash's difficulty retargeting behaves differently from Bitcoin's, both in window length and in how sharply it responds to hashrate variance. Shorter effective windows mean difficulty tracks hashpower more responsively, which in a period of rapid fleet turnover produces a jagged, staircase-like difficulty chart. Read a one-year chart of Zcash difficulty next to Bitcoin's and the difference is visible to the naked eye. The report's 2.5x figure is real, but it is drawn from a curve that oscillates, and oscillation is a signature of short-horizon capital, not long-horizon conviction.
The yield math, and the electricity assumption hiding inside it
Here is where the analysts among you should start paying attention.
The report's central quantitative claim rests on an electricity cost of $0.05 per kilowatt-hour. That is the single most load-bearing assumption in the entire document, and it is stated almost in passing.
Five cents is not a made-up number. It is achievable — in specific places, under specific conditions. Stranded natural gas in the Permian Basin. Certain Ethiopian and Paraguayan hydro sites. Kazakh coal-adjacent capacity during off-peak windows. Iranian industrial rates, at least nominally. A handful of legacy Chinese hydro contracts that never fully unwound after the 2021 migration. In each case the miner is monetizing a power source that has no better buyer, and the true economic cost is closer to the opportunity cost of that stranded resource than to any published retail rate.
But five cents is not representative. The U.S. industrial average sits somewhere between twelve and sixteen cents. Most of Western Europe is north of twenty, with Germany and the United Kingdom considerably higher once grid fees are loaded in. A miner paying eighteen cents is not earning four times Bitcoin's yield per megawatt-hour. That miner is underwater on Zcash and comfortably profitable on Bitcoin, because Bitcoin's machines are more efficient per unit of work even when their per-unit reward is lower.
I have made this exact mistake myself. In 2017, I spent three weeks modeling Golem's reward distribution against its whitepaper claims, and the flaw I found was precisely this shape: a mechanism that looked sound at the assumed parameter values and collapsed at realistic ones. The whitepaper had priced in an idealized fee environment. Real fee environments are volatile, and volatility is where the model breaks.
Same pattern here. The Zcash yield advantage is real. It is also conditional. The conditions are power price and machine efficiency, and both are moving.
Let me put a rough number on the sensitivity, because hand-waving is how people lose money. Take an Antminer Z15 Pro at its rated throughput and a 2,650-watt draw, running at a five-cent power cost. Power alone runs you a little over three dollars a day. At the report's implied reward rate, gross daily revenue clears that comfortably — which is what produces the headline ratio. Now re-run it at twelve cents. Power cost more than doubles. Re-run it at twenty and the machine is burning money the moment difficulty ticks upward.
This is not an argument that the report is wrong. It is an argument that the report's conclusion is geographically bounded, and the geography is not stated in the headline.
Where the reward actually comes from
Now the structural part, which I find more interesting than the yield.
Zcash has a hard cap of 21 million coins and a halving schedule modeled on Bitcoin's. Every coin that has ever been mined arrived through the block subsidy. There is no protocol-level fee capture, no burn mechanism, no staking yield, no MEV, no sequencer revenue. One hundred percent of miner income is new issuance.
The early distribution carried a Founders Reward — a percentage of block rewards directed to the Electric Coin Company, the Zcash Foundation, and early investors — which decayed over time and was the subject of years of community argument before being restructured into a revised development fund allocation at the 2024 halving. That history matters less for its numbers than for its precedent: Zcash has demonstrated that it will renegotiate its own issuance policy through public dispute. In a proof-of-work network, that is an unusual degree of governance flexibility, and it cuts both ways. It means the rules can adapt. It also means the rules are not fixed.
Compare that to the networks I spend most of my time on, where token holders argue endlessly about real yield and value accrual. Zcash has none of it, and in a strange way that is cleaner. There is no pretense. The coin is a payments and privacy instrument, and its security budget is a pure function of issuance multiplied by price.
That purity has a cost. When I wrote The Yield Trap during DeFi Summer in 2020, my argument was that high annual percentage yields were masking systemic liquidity risk — that the number on the screen was a price signal dressed up as a fundamental. The Zcash situation is the same shape in a different costume. High mining profit is not a property of Zcash. It is a property of ZEC's price relative to its network difficulty at a given moment, and it will mean-revert exactly as fast as those two variables allow.
The report describes a reinforcing loop: price rises, mining becomes profitable, hashrate grows, network security improves, confidence rises, price rises further. That is a real dynamic. I have watched it operate in Bitcoin for over a decade.
But a reinforcing loop is not a foundation. It is a feedback path. And feedback paths run in both directions.
The security budget nobody wants to talk about
Here is the invariant I keep coming back to.
In the chaos, look for the invariant. For a proof-of-work chain, the invariant is the cost of attack. Roughly: how much does it take to acquire enough hashrate to reorganize the chain or double-spend?
For Bitcoin, that number is now in the tens of billions of dollars at spot hardware prices, and the network spends $35 million a day to defend it. For Zcash, daily issuance is about $2 million — roughly 5.7 percent of Bitcoin's — and the hashrate base is orders of magnitude smaller. A 2.5x increase from a low base leaves you with a low base multiplied by 2.5. That is not a rhetorical trick. It is just what the exponent does.
I am not arguing Zcash is insecure. Its security model has held for eight years, and a $2 million daily budget is enough to deter the classes of attacker who target small chains for opportunistic reorgs. But it is not enough to deter a state-level or well-capitalized adversarial actor — and neither, strictly speaking, is Bitcoin's, though the difference is one of degree, and the degree is enormous.
The point is that hashrate up 2.5x sounds like security up 2.5x. It is not. Security scales with the cost of acquiring the marginal hash, which scales with hardware availability and energy contracts, not with a percentage change in a dashboard metric. If the incremental hashrate came from existing fleets being pointed at Zcash — the substitution case — then the same machines can be pointed somewhere else within the hour, and the security they provided was rented, not owned.
This is the sentence the report should have written and did not: the hashrate that arrives cheapest is the hashrate that leaves first.
The compliance hedge, and why Grayscale matters more than the yield
Which brings me to what I think is the actual story, buried under the mining economics.
Zcash has always had a split personality. Its shielded addresses provide genuine transactional privacy. Its transparent addresses are, for all practical purposes, Bitcoin. And critically, the protocol supports selective disclosure through viewing keys — a mechanism by which a shielded transaction's contents can be revealed to a designated third party, such as an auditor or a regulator, without exposing anything to the public.
That capability is not a bug fix. It is a strategic posture, and it is the same posture PayPal adopted when it launched PYUSD. Both are bets that the winning move against hostile regulation is not resistance but pre-emption — become legible to the regulator before the regulator decides to make you illegible.
I have written about this dynamic before, and I will keep writing about it because it keeps being right. The pattern in U.S. crypto regulation has not been ignorance of the technology. It has been deliberate ambiguity — enforcement as a substitute for rulemaking, where the rule is whatever the last settlement said it was. A privacy asset cannot win a fight against ambiguity. It can only win by removing the ambiguity on its own terms, which means shipping the compliance primitives before anyone demands them.
The pressure is not only American. Europe's AML framework has tightened around anonymous instruments, several jurisdictions have narrowed exchange access to privacy assets, and the practical result is that a privacy coin's survival depends less on its cryptography than on whether regulated intermediaries can describe it to a supervisor without flinching. Selective disclosure is the answer to that question. It is also the only answer the protocol has.
Grayscale publishing a favorable research note on Zcash mining is not, in itself, a compliance event. But Grayscale is a regulated asset manager, and its research output is a leading indicator of what it intends to package. A trust product requires a legal theory. A legal theory requires the asset to be describable to a regulator in terms the regulator accepts. The report's emphasis on mining economics rather than privacy — on yield, on hashrate, on energy efficiency — is not an oversight. It is a framing that makes Zcash legible as an industrial asset rather than as a privacy instrument.
Read the report as a mining analysis and you will learn something about Equihash. Read it as a positioning document and you will learn something about what happens next.
The contrarian read
Everything above is consistent with the report. Here is where I part ways with it.
The report's implicit thesis is that Zcash mining offers a superior risk-adjusted return relative to Bitcoin mining, and that this gap represents an inefficiency. I think the gap is real, and I think it is not an inefficiency. It is a risk premium, and the market is pricing something the report does not name.
Call it the delisting premium.
Privacy coins have been progressively squeezed out of regulated venues over recent years. Exchange availability has narrowed in several jurisdictions, and the trend has been one-directional. If ZEC is harder to sell than BTC in the venues where institutional capital actually operates, then ZEC's market depth is structurally thinner, its slippage higher, and the realized price a miner receives lower than the quoted spot. A yield advantage that survives on paper can evaporate in execution. The report compares machine-level revenue in the abstract. It does not compare executable liquidity. That is the gap between a model and a market.
There is a second omission, and I consider it more serious. The report presents hashrate growth as evidence of network health without presenting a single user-side metric. No active address count. No shielded pool share. No transaction volume trend. I went looking for a correlation between Zcash hashrate and on-chain utilization across the last three years, and I could not find one that holds. Hashrate tracks price. It does not track adoption. Those are different curves, and conflating them is how mining narratives get sold to people who do not mine.
Solitude is the price of clear vision. The clear vision here is unglamorous: Zcash is a technically mature, eight-year-old proof-of-work network with a real privacy primitive, a thin liquidity profile, a single-vendor ASIC dependency, a security budget small in absolute terms, and a mining economy that is profitable primarily where electricity is cheapest and primarily because a price move made it so.
That is not a bearish case. It is a description. The bearish case requires you to believe the loop reverses. The bullish case requires you to believe the loop compounds. Both are bets on ZEC's price wearing the costume of a mining analysis.
Math does not care about your conviction. It only cares about your inputs.
What I am watching
Three signals, ordered by how much they would change my model.
First, the ZEC-to-BTC ratio plotted against the network difficulty curve. If price holds while difficulty keeps climbing, the yield advantage compresses mechanically and the marginal miner exits. If price rises faster than difficulty, the loop extends. The spread between those two lines is the whole trade, and it is observable in real time by anyone willing to look at it rather than read about it.
Second, whether Grayscale moves from research to filing. A trust application would be the first hard evidence that the compliance posture I described is being operationalized. It would also force a regulatory response, and the shape of that response — clarity or continued ambiguity — would tell you more about Zcash's next five years than any hashrate chart. Pay attention to how the filing describes the asset. If the words privacy and anonymity appear sparingly and the words industrial and infrastructure appear often, you will know exactly which door the industry has chosen to walk through.
Third, the shielded pool share. If the privacy primitive is the actual product, that number should trend up over time regardless of price. If it stalls while hashrate grows, then the network's growth is purely financial and the privacy narrative is decorative. I would also watch the developer surface. There was a time when the roadmap toward cross-chain shielded swaps and even a longer-horizon discussion about shifting the consensus mechanism entirely seemed like the natural next chapter. Roadmaps are the cheapest thing a protocol produces. Shipping is the expensive part, and Zcash's expensive part has been quiet lately.
I have watched three narrative cycles turn over in this market — the ICO era, DeFi Summer, and the institutional alignment that followed the spot ETF approvals. Each one looked like a new paradigm at the peak and like a feedback loop in retrospect. The Zcash mining story is smaller than any of them. But it has the same skeleton: a real mechanism, a real constraint, and a headline that quotes the mechanism while omitting the constraint.
The question worth asking is not whether Zcash mining is more profitable than Bitcoin mining today. It is what the network has to become for that to still be true in eighteen months — and whether anyone currently pointing machines at it has thought about the answer. Most of them, I suspect, are renting the future one difficulty adjustment at a time.
