The Table Determines the Outcome
Derek Wayne Bailey9 min read·Just now--
How photonic computing rebuilds Bitcoin mining at the substrate.
At a Manhattan steakhouse last quarter, a family-office principal opened a slim folder and walked his table through the math on a Bitcoin mining contract he was about to sign. Two-point-eight million dollars for three hundred next-generation Antminer S21 Pros, racked at a hosted facility in Quebec. Six-cent power. Twenty percent management fee. Daily BTC payouts to a wallet. The math, drawn by the seller on the back of a steak menu, returned the principal in roughly twenty-four months. The folder was passed across the table. A second principal, who had run a similar deal eighteen months earlier and was now sitting on machines that had lost forty percent of their resale value, said the words every person in the industry has heard a thousand times.
It pencils.
It also will not matter.
The container as a category.
The high-net-worth turnkey Bitcoin mining market in 2026 is a category roughly fifteen years old. It sells one product wearing four labels. The first is Buy and Host, where the client purchases the ASICs and the provider racks them at a managed colocation site. The second is Fully Managed Mining, an eighty-twenty profit split with the operator running the machines under a custody account on the client’s behalf. The third is the Turnkey Container, a one-megawatt steel box dropped on a power site preloaded with about three hundred next-generation miners. The fourth is the Guaranteed Hashrate subscription, a one-time fee in exchange for a contracted terahash delivery over thirty-six months. Underneath the four labels, the offer is the same. The client is buying transistor-based ASICs and the electricity to run them, with a provider doing the racking and the remote-hands work in the middle.
The current benchmark machine is the Antminer S21 Pro at 234 terahashes per second and 15 joules per terahash, with the S21 XP Hyd above it at 473 terahashes per second and 12 joules per terahash. Hosting runs six to ten cents per kilowatt-hour. Siting is dictated by stranded power and cool climate — Texas, North Dakota, Nebraska, Washington, Alberta, Quebec — because the binding constraint on the entire industry is the cost of feeding electrons through transistors fast enough to do useful work.
Every line in every contract eventually reduces to two costs. The first is the price of a transistor-based machine, which depreciates on an eighteen-to-twenty-four-month cycle as each new generation halves the residual value of the prior one. The second is the lifetime electricity bill, which under current network difficulty and Bitcoin price determines whether any given operation is profitable in any given month. The whole category, examined honestly, is a power-and-depreciation business with a Bitcoin coupon attached.
The error nobody questioned.
SHA-256 is a logical operation. It does not require silicon, copper, or heat. It requires a logic element capable of executing the hash function and a path to deliver the result. The transistor was the only logic element available when Bitcoin was designed in 2008, and the entire global mining industry has since organized itself around the energetic cost of running that logic element at industrial scale.
Every joule per terahash quoted on a spec sheet is a measurement of how much heat the transistor produces while doing the work it was not designed to do efficiently. Every megawatt of imported power, every gallon of cooling water, every cubic foot of refrigerated air pushed through a hosted facility, is a tax on running cryptographic logic on a substrate that was never optimized for it. The industry calls this efficiency. It is the residue of a substrate choice nobody questioned, because for fifteen years there was no alternative substrate to question.
This is the trap the high-net-worth buyer is walking into when they sign the contract on the steak menu. They are not buying hashrate. They are buying a power bill, on a depreciating machine, on a substrate the industry has spent fifteen years optimizing not because the optimization is meaningful, but because no other substrate existed.
Light has its own logic.
There is now another substrate on the near-term horizon. Photonic logic — the execution of Boolean and cryptographic operations in light rather than in electron flow — has cleared its validation gates. The Poovey Switch, a photonic logic element measured independently at the Technion Israel Institute of Technology under Report #9105027635, switches at 150 to 200 femtoseconds, against a prior-literature best of 600. True Photonic, Inc. has filed a Hash Engine provisional patent in January 2026 directed specifically at SHA-256 and related cryptographic primitives, sitting inside a portfolio of nineteen application patents and roughly eight hundred claims on the Logic-in-Light™ stack.
The same logical operation, executed in light, draws on the order of ten percent of the energy of the transistor-based equivalent. It produces no resistive heat that requires water or industrial air handling to remove. It runs at switch speeds three orders of magnitude faster than the clocked-transistor equivalent. None of that is a refinement of the existing industry. It is a substrate shift, of the kind that has happened twice before in the history of computing — vacuum tubes to transistors in the 1950s, and discrete transistors to integrated circuits in the 1960s. Each of those shifts ended an entire generation of incumbent equipment, not because the incumbent equipment got worse, but because the underlying substrate got cheaper, faster, and cooler than anything the prior substrate could become.
What collapses with the heat.
When the heat goes, four things go with it.
The first is the power bill. The benchmark Antminer S21 Pro burns fifteen joules to compute a single terahash of SHA-256 work, the headline number on its spec sheet; field-deployed fleets average somewhere between seventeen and twenty-five. The photonic equivalent is designed to deliver the same terahash at roughly one tenth that energy. Multiply the delta across the trillions of hashes a facility computes per second, and the cost that defines whether mining is profitable in any given month — the power bill — stops being the binding constraint on the entire industry. The grid stops being the bottleneck; a rooftop’s worth of solar replaces a substation.
The second is the cooling water. Not less. Zero. Light produces no resistive heat to remove. The legacy industry pays for cooling twice — between $150,000 and $400,000 of capital per megawatt for air-cooled facilities, and $280,000 to $600,000 per megawatt for immersion systems, before another ten to twenty percent of the lifetime electricity bill is burned just to move heat around the building. On the photonic substrate, both line items disappear, because there is nothing to immerse and nothing to cool.
The third is the container itself. One photonic processor, executing SHA-256 hash logic in light, replaces three hundred to four hundred ASIC miners. The unit of growth in the new industry is not the megawatt and not the container. It is the pod — a kilowatt-class footprint inside a Clean Compute Center that produces the same hashrate the legacy industry produces inside a steel box on stranded grid power.
The fourth is the hardware obsolescence cycle. The legacy industry’s eighteen-to-twenty-four-month refresh treadmill is a property of transistor lithography, not of cryptographic computing. On the photonic substrate, the logic element is not lithographically committed to a single hash algorithm. The same reserved capacity that mines Bitcoin today can be allocated to Scrypt, Ethash variants, Equihash, or any other proof-of-work workload as profitability shifts, without changing hardware. The machine does not become obsolete. The contract carries forward.
The HNW market is the first to crack.
There is a reason the high-net-worth turnkey market will be the first segment of the legacy industry to feel the displacement, and the reason has nothing to do with technology adoption curves. It has to do with how high-net-worth capital prices total return.
The institutional miner — the public Bitcoin mining company with shareholders and grid contracts and quarterly guidance — is locked into a multi-year capital plan. Their machines are on order. Their power contracts are signed. Their analysts have already modeled the next two halvings. They cannot pivot without writing down assets, restating guidance, and explaining to the board why the substrate they bought eighteen months ago is being replaced by a substrate whose Hash Engine patent was filed last quarter. The institutional miner will be the last to move.
The high-net-worth buyer has none of those constraints. The high-net-worth buyer is sitting at a steakhouse with a folder, and the question they are asking is not what does my CFO think but what does this cost and what does it return. When the math on the steak menu changes — when the same hashrate becomes available at a tenth of the power bill, with no hardware depreciation, with multi-coin assignment rights, on a contract that does not require a megawatt of stranded power and does not lose forty percent of its asset value in eighteen months — the high-net-worth buyer will move first. They are not optimizing for industry continuity. They are optimizing for return.
The market is also where the legacy industry is most exposed. The four-label turnkey product is a power-and-depreciation business with a Bitcoin coupon attached, and the high-net-worth buyer has historically priced both the power and the depreciation into the deal. When a substrate shift removes both of those costs, the legacy product loses what made it pencil. The folder closes.
The window measured in months.
The substrate shift will not happen on a schedule the legacy industry can manage. It will happen on the schedule the photonic stack ships. The first photonic-electronic pluggable products, which deliver an immediate efficiency improvement on the order of thirty percent over a pure transistor-based fleet, are entering qualification now. The native two-inch photonic pluggable, which delivers the full hundred-fold density ratio against ASIC miners, follows on the staged timeline of any genuinely new substrate. Both will be in market well within the depreciation window of the Antminer S21 Pro that the principal at the steakhouse is about to buy.
By the time his ASIC contract has paid back its capital, the substrate underneath it will have been replaced by something that runs at a tenth the power on a pod-class footprint with no cooling water and no hardware refresh treadmill. His machines will still hash. They will hash on a network where the marginal cost of hashrate has collapsed, and where the buyers who held back, or who exited the legacy product before the displacement priced in, will be holding the new substrate.
This is not a story about Bitcoin going away. Bitcoin will still settle blocks every ten minutes for as long as the network operates. This is a story about who mines Bitcoin, on what substrate, at what cost. The next era of Bitcoin mining will not be the megawatt era. It will be the photon era. The transition is already underway, and the high-net-worth turnkey market is where the transition will be visible first, because high-net-worth capital has the fewest reasons to stay loyal to the megawatt.
What comes after the transistor.
What comes after the transistor in cryptographic computing is not an incrementally better transistor. It is a different logic element. The patents have been filed. The substrate has been validated. The economics, when worked through honestly on either side of a steak menu, point in one direction. The high-net-worth buyer who sees this clearly today will do what high-net-worth buyers have always done at the end of an asset cycle: they will price the displacement before the market prices it for them.
While that turnkey ASIC miner deal folder is being passed around the table, another folder is about to be passed at a second table. Same steakhouse. But this folder has photonic miners.
Different substrate. Different math. A different outcome.
Derek out.
Derek W. Bailey is the Founder of True Photonic, Inc. and the author of Keep Computing: How Light Solves Computing’s Impossible Problem (True Photonic, 2026). He publishes on Substack and Medium under @bailey55015. Contact: [email protected].
Disclosure: The author has a financial interest in the success of photonic computing technology. Performance projections are based on preliminary research, early-stage testing, and technical simulations. Actual results may vary.
Forward-Looking Statements: This article contains forward-looking statements regarding future technological capabilities and market developments. These statements involve risks and uncertainties that may cause actual results to differ materially. Readers should not place undue reliance on forward-looking statements, which speak only as of the publication date.
