On September 9, a Grayscale research note circulated through institutional inboxes with a number that should have been boring and was not. The Zcash network's hashrate had grown 2.5x year over year.
That figure is a confession. Equihash, the proof-of-work algorithm Zcash has run since its October 2016 genesis block, was engineered as a memory-hard function specifically so that commodity GPUs could compete and application-specific integrated circuits could not. Ten years later, a 2.5x hashrate expansion in twelve months is not the signature of hobbyists plugging spare rigs into garage outlets. It is the signature of a supply chain: Bitmain's Antminer Z15 Pro, shipped in volume, racked in 2,650-watt increments, amortized across twenty-four-month depreciation schedules. The network did not become more decentralized. It became more industrial.
The second number in the same note is stranger. A Zcash miner, Grayscale's analysts calculated, earns roughly twice the daily revenue of an equivalent Bitcoin miner per machine, and roughly four times the revenue per megawatt-hour of electricity consumed. The whole Zcash network pays out about $2 million a day in block rewards. Bitcoin pays out about $35 million. On less than six percent of Bitcoin's security budget, Zcash is quoting more than double the per-unit return.
Two numbers pulling in opposite directions. One describes industrial consolidation. The other describes a yield premium. Both are accurate. Neither is the whole story, and the gap between them is where the actual analysis lives.
Zcash occupies a specific and uncomfortable position in the crypto taxonomy. It is a Layer 1 payment network with an optional privacy layer, launched by Zooko Wilcox and the Electric Coin Company, and it was the first production deployment of zk-SNARKs โ zero-knowledge succinct non-interactive arguments of knowledge โ in a live monetary system. The technical lineage runs from Sprout, the original 2016 shielded pool with its heavyweight trusted setup ceremony, through Sapling in 2018, which cut proof generation from minutes to seconds, to Orchard under network upgrade NU5 in 2022, which eliminated the trusted setup entirely through the Halo 2 proving system.
That progression matters for reasons the mining-yield discussion ignores. Halo 2 removed the single most-criticized structural weakness in Zcash's design โ the requirement that a ceremony's participants destroy their toxic waste or else hold a silent minting key over the entire supply. A chain that once asked you to trust a group of strangers to throw away a laptop file now asks you to trust mathematics. Very few networks have completed that transition, and none did it first.
Equihash itself is a generalized birthday problem with tunable parameters โ Zcash runs at (n=200, k=9). The memory-hardness came from the fact that solving it efficiently requires holding a large working set in RAM, which in 2016 made custom silicon economically pointless. The thesis was elegant: if the algorithm rewards memory bandwidth rather than raw computation, then a gamer's graphics card competes with a purpose-built chip, and issuance stays distributed.
The thesis died in 2018. Bitmain shipped the Antminer Z9 mini, and every generation since has widened the gap between ASIC and GPU economics. This is a general law rather than a Zcash-specific failure. Any proof-of-work algorithm with a liquid price attached to its reward will be industrialized, because the reward is a cash flow and cash flows attract capital expenditure. Memory-hardness did not prevent ASICs. It only delayed them by two years and raised the capital cost of entry.
Zcash's governance sits in the same institutionally awkward middle. Supply is capped at 21 million ZEC with a roughly four-year halving schedule โ the second halving landed in November 2024, cutting the block subsidy from 3.125 to 1.5625 ZEC. Twenty percent of that subsidy has historically been redirected to a development fund split among the Electric Coin Company, the Zcash Foundation, and the Zcash Community Grants committee, a structure reconfigured under network upgrade NU6. This is an inflation tax on holders levied to pay developers, and it remains one of the more honest funding models in the industry precisely because it is explicit rather than disguised as a treasury.
What Zcash is not matters equally. There are no smart contracts. There is no staking. There is no DeFi lending market, no liquid staking derivative, no restaking layer. It is a payment rail with an optional privacy mode, competing against a mandatory-privacy incumbent in Monero and against the gravitational pull of Bitcoin's network effects. Understanding the mining economics means understanding that the entire value of the network flows through one channel: the price of a coin that people use to move money without being observed.
The per-megawatt-hour comparison is the only honest unit for evaluating mining economics, because mining is fundamentally a business of joules. Machines are just the vessel. Grayscale's four-times figure, however, embeds an assumption that the note states plainly and then treats as neutral: electricity at $0.05 per kilowatt-hour.
Run the arithmetic on the marginal operator. An Antminer Z15 Pro draws roughly 2,650 watts, which is 2.65 kilowatts, which across a thirty-day month at continuous load is approximately 1,908 kilowatt-hours. At five cents, that is $95 of power per machine per month. At twelve cents โ roughly the Austrian and German industrial rate my own colleagues in Vienna pay, and close to the US commercial average โ it is $229. At twenty cents, which is what a small operator in New England or coastal California actually faces after demand charges and transmission fees, it is $382.
Now put revenue against that ladder. The note's implied gross revenue per Z15 Pro sits in the range of roughly $360 to $450 a month at the prices it used. Subtract $95 of power and you have a fat margin. Subtract $382 and the margin is thin enough that a single difficulty adjustment erases it. The premium that was four times becomes something closer to 1.2 times, and any operator paying retail rates is running a business that exists at the pleasure of the difficulty formula.
Mining margins are not a property of the protocol. They are a property of the cheapest electricity a marginal operator can contract, and they belong to whoever holds that contract.
The corollary is that a visible yield premium is a magnet. Capital flows toward the arbitrage until the arbitrage closes. A 2.5x hashrate expansion is precisely what that process looks like in motion: miners read the same note, ordered the same hardware, and are now competing away the spread. Per-machine revenue falls roughly in proportion to hashrate growth holding price constant โ a 2.5x hashrate increase cuts each machine's share of the $2 million daily reward pool by about sixty percent. The Grayscale numbers describe a window, not a plateau.
So what does the 2.5x actually tell us, if not that mining is permanently lucrative? It tells us that a specific population of capital allocators expects Zcash to appreciate. Mining is a forward contract on price with an electricity-denominated strike price. When you buy an ASIC, you are buying a levered, illiquid, two-year-duration call option on the underlying asset, financed by a power bill.
Hashrate growth is a pricing signal in disguise โ a delta-one bet on the underlying, sized by the cost of power available to the buyer.
This matters because the reflexive loop Grayscale describes is real, and it is also ungoverned. The loop runs: price rises, mining margin expands, hashrate grows, nominal security budget rises, confidence improves, price rises. Every step reinforces the next. It is a beautiful machine for as long as it turns in one direction.
I watched a structurally identical loop in 2022, when Terra's mechanism promised that demand for UST would burn LUNA, which would raise LUNA's price, which would attract more demand for UST. The mechanism worked exactly as documented. It simply had no brake, and the reversal was not a correction but a cascade. I spent that spring on a five-person audit of our own DAO's treasury, rebalancing collateral positions across Aave and Compound while total value locked across the ecosystem fell forty percent in a matter of days, and what I learned is that reflexive systems do not fail gradually. They fail when the marginal participant changes direction.
Crisis is just code with a high gas fee. The fee is paid by everyone who assumed the loop only spun one way.
For Zcash, the reverse loop is specifically corrosive to the thing the network sells. If ZEC's price falls forty percent, the marginal operator โ the one paying twelve cents or more per kilowatt-hour โ switches off. Hashrate falls. Nominal security budget contracts. And for a chain whose entire proposition is that your transaction cannot be read, the second-order implication is that your transaction can be reorganized away. Privacy without finality is a log file with a nicer interface.
That brings the analysis to the number nobody quoted in the marketing summary: the security budget ratio. Zcash pays roughly $2 million a day to secure itself, which annualizes to about $730 million. Against a market capitalization hovering in the high hundreds of millions of dollars, that is a ratio approaching one to one โ the network spends roughly a dollar a year on security for every dollar of value it claims to protect.
Bitcoin, by comparison, pays about $35 million a day, or roughly $12.8 billion a year, against a market capitalization in the trillion-plus range. That ratio is closer to one to a hundred.
Zcash spends on the order of a hundred times more of its market value on security than Bitcoin does. That is either the most honest security budget in crypto, or the most fragile.
It is honest because it reflects a network that has not inflated its own valuation narrative. It is fragile because the absolute number is small enough to matter. A 51% attack on a proof-of-work chain costs the attacker the price of acquiring or renting enough hashrate to outpace the honest chain for the required window. The Equihash ASIC market is thin, concentrated, and illiquid โ which cuts both ways. It means renting an attack is harder than on a chain with a deep hash-rental market, but it also means a single buyer with capital and a hardware relationship could accumulate a dangerous share quietly, off-exchange, without the visibility that an Ethereum staking queue provides.
And an attacker who controls an Equihash majority can do something an attacker on a transparent chain cannot: censor transactions they are unable to read. They cannot decrypt shielded notes. They can simply refuse to include them. For a privacy network, exclusion is the attack, not surveillance.
The Grayscale note frames all of this as a mining yield story. That is the wrong frame, and it is the frame a distribution business would naturally choose. Mining yield on a proof-of-work chain is a derivative of the underlying asset's adoption. It is the symptom, not the cause.
The adoption data is where the privacy premium gets tested, and this is the part of the analysis that the mining math never touches. Zcash has two transaction modes. Transparent addresses work like Bitcoin addresses โ fully visible, fully traceable, fully compatible with every exchange's compliance pipeline. Shielded addresses hide sender, receiver, and amount inside a zero-knowledge proof. The network allows users to choose, and users overwhelmingly chose transparent for most of Zcash's history.
The fully shielded pool has historically held a low single-digit to low double-digit percentage of circulating supply. Transaction-level shielding rates tell a similar story, with shielded activity rising meaningfully only in recent years as Orchard made private transactions cheaper and mobile wallets matured. Even accounting for that improvement, the honest characterization is that Zcash is a transparent chain with a privacy feature, not a privacy chain with a transparency concession.
A privacy coin whose shielded pool is a minority of supply is a transparent coin with an optional feature โ and options decay when nobody exercises them.
Here is why that matters more than any difficulty chart. The mining premium Grayscale identified is a market's valuation of Zcash's scarcity, its halving schedule, and its halving-driven supply curve. It is not a valuation of privacy demand, because privacy demand does not show up in the revenue line. Block rewards are paid in newly issued ZEC regardless of whether anyone is using the shielded pool. A miner earning $400 a month from a Z15 Pro has no idea and no exposure to whether those coins are being spent privately. The revenue is identical if shielded usage is two percent or eighty percent.
That decoupling is the deepest structural weakness in the bull case. The report's own logic โ high margin, hashrate growth, security improvement, price confidence โ describes a self-referential loop that never once requires an end user to need privacy. The market can price Zcash's scarcity without ever pricing its utility. That is not a durable foundation. It is a financing round that happens to be paid in electricity.
Monero is the comparison that clarifies the choice. It mandates privacy on every transaction, which produces a coherent product and a hostile regulatory posture โ delisted from major exchanges, functionally excluded from institutional custody, and unable to build a compliance bridge. Zcash chose the opposite trade: an optional privacy layer with viewing keys that let holders disclose their transactions voluntarily to auditors, exchanges, and tax authorities. That design decision is why a regulated asset manager can publish research on it at all.
This is where the regulatory layer stops being background noise and becomes the load-bearing wall. In August 2022, the US Treasury's Office of Foreign Assets Control sanctioned Tornado Cash, a set of immutable smart contracts, and in doing so implied that publishing code could constitute a sanctionable act. In November 2024, the Fifth Circuit ruled in Van Loon v. Treasury that OFAC had exceeded its statutory authority, because immutable smart contracts are not property that can be sanctioned. OFAC delisted Tornado Cash in March 2025. The civil question was answered.
The criminal question was not. The prosecutions of the developers who wrote that code continued, and the theory behind them โ that maintaining software can constitute operating an unlicensed money transmitting business โ remains live. That theory does not care whether a chain is optional-privacy or mandatory-privacy. It targets the act of writing. Every open-source developer building financial infrastructure is downstream of how that question resolves.
I spent 2024 inside the Austrian implementation of MiCA, working with a Vienna policy think tank, organizing three town halls that put developers and legal experts in the same room with two hundred people each. Our argument was narrow and technical rather than ideological: that zero-knowledge proofs can satisfy compliance obligations without destroying user sovereignty, because a proof of solvency or a selective disclosure of origin can be verified without revealing the underlying data. We amended two minor clauses in the local draft. Two clauses is not a victory you put on a poster. It is a victory you file and build on.
For a privacy chain, regulatory clarity is not a constraint on the product. It is the product's distribution channel.
Everything about Zcash's institutional trajectory follows from that. The Grayscale note is not neutral research. Grayscale is a regulated asset manager with a distribution business, and research is upstream of product. Look at how Bitcoin's ETF approval reshaped the asset: custody rails, authorized participants, and a price increasingly set by allocators rebalancing quarterly mandates rather than by individuals choosing a payment method. The peer-to-peer electronic cash thesis did not survive contact with the ETF wrapper. The wrapper won.
The same trajectory is available to Zcash, and the mining math is the opening argument for it. When a regulated manager publishes favorable economics on a privacy asset, the note is doing three things at once: explaining a yield, normalizing the asset class, and testing institutional appetite for a wrapper. The protocol remembers what the regulators forget โ that privacy is a requirement for commercial activity, not a perversion of it. Enterprises do not broadcast their payroll. Funds do not publish their positions. The absence of privacy in traditional finance is not a design preference; it is a surveillance subsidy that everyone has quietly agreed to stop questioning.
The transmission chain from this note to actual value is short and narrow. Upstream, Bitmain and the small set of Equihash ASIC manufacturers capture hardware demand โ and the Z15 Pro's dominance means a single vendor's supply chain is the constraint on the network's security growth. Midstream, mining farms with stranded or subsidized power capture the arbitrage, and they capture it in proportion to how far below twelve cents their electricity actually costs. Downstream, exchanges and custody providers capture whatever flow materializes from renewed institutional interest.
The holder of ZEC captures two things. First, better security in nominal terms, which is real but priced against a small absolute base. Second, dilution, since the entire $2 million daily reward is newly issued supply sold into the market to cover operating costs. There is no protocol fee, no buyback, no burn. Zcash's token model is a pure halving curve with a development tax attached.
There is also a hardware concentration risk that the yield math obscures. Equihash ASIC production is a rounding error in Bitmain's wafer allocation. Bitcoin ASICs command far larger margins per unit of fabrication capacity. If Bitcoin's price rises faster than Zcash's, the foundry slots go to SHA-256 hardware and Equihash rigs sit in a backlog. A single-model dependency means the difference between a 2.5x hashrate year and a flat one may be decided in a chip allocation meeting in Shenzhen, not in the Zcash price chart.
Open source is a promise, not a product. Zcash's code is verifiable by anyone with the skills to read it. Its development is funded by an inflation tax that holders pay involuntarily and can exit only by selling. That is a defensible structure. It is not a structure that scales indefinitely when the market capitalization it taxes is smaller than the annual budget of a mid-sized university.
The consensus reading of the Grayscale note will be simple: Zcash mining pays better than Bitcoin mining, therefore buy ZEC, therefore mine ZEC. The contrarian reading inverts the causality. Mining profitability is a lagging indicator of price, and price in a small-cap asset is a leading indicator of narrative rather than adoption. The premium exists because ZEC has outperformed BTC on the margin; it does not exist because more people are shielding transactions. If ZEC underperforms for two consecutive quarters, the same report writes itself backwards, and the hashrate that arrived in twelve months can leave in six weeks. Machines can be unplugged faster than they can be installed.
The second blind spot is jurisdictional. Censorship-resistant networks with cheap-power economics attract hashrate from wherever electricity is cheapest and enforcement is thinnest. That is a genuine feature โ it makes the network expensive to censor politically โ and a genuine risk, because the same property makes the network's security budget sensitive to policy changes in jurisdictions that are not party to any monetary negotiation. There is no committee controlling who points an Equihash ASIC at the chain. There never will be.
The third blind spot is the one that determines whether any of this matters in five years. What would actually re-rate Zcash is not a mining premium. It is the day shielded transactions become the default rather than the option โ when the wallet software makes private the path of least resistance and transparent the deliberate exception. That is a product decision, executed by a small team, measurable in pool composition data, and entirely independent of how many Z15 Pros are hashing. Watch the shielded pool share, not the difficulty chart.
Speed without direction is just volatility. A hashrate chart going up and to the right looks like progress. It is only progress if the chain it protects has something worth attacking.
Three signals deserve attention over the next four quarters, and none of them is the hashrate headline. The first is the shielded pool's share of circulating supply, which is the only unambiguous measure of whether privacy is being used rather than sold. The second is difficulty-adjusted per-machine revenue at a realistic power cost, because the four-times premium is a function of an assumption many operators cannot access. The third is whether the regulated distribution channel actually opens โ a ZEC vehicle filing, a custody integration, a custody-bank disclosure โ because that is what converts institutional attention into institutional flow.
Regulation is the friction that forces efficiency. The friction is arriving. The efficiency has not yet been demonstrated. Zcash has spent nine years proving that privacy can be built, and roughly two years proving that it can be described to a regulator without being dissolved. What remains unresolved is whether it can be used. The mining math measures the price of that question. It cannot answer it.