Lucas runs a hosted fleet out of a converted warehouse outside Sheridan, Wyoming, and in March he did something that would have been physically impossible two years ago. He pulled a half rack of aging Antminer S19 units, forty machines, out of a cabinet, and slid a single chassis into one 3U slot to replace all of them. The new unit produced more hashrate than the forty it replaced. It drew power through one connector instead of forty. It reported to his dashboard as one endpoint. That chassis was the Bitmain Antminer S23 Hyd 3U, and it is the first production Bitcoin miner to cross the petahash barrier inside a single box.
At 1,160 TH/s (1.16 PH/s) of SHA-256 hashrate, 11,020 watts of draw, and 9.5 joules per terahash of efficiency, the S23 Hyd 3U sits at the top of Bitmain's current lineup. Those headline numbers get repeated across every spec sheet and distributor page. What almost nobody explains is the part that actually decides whether you should buy it: the density premium. The 3U and a pair of standalone S23 Hydro units deliver nearly identical aggregate hashrate at identical efficiency. So the real question is not whether this machine is powerful. It is whether you pay extra per terahash for management consolidation, and at what fleet size that premium starts paying you back.
This review works through the verified specs as of June 1, 2026, the real profitability at four electricity rates, the strictly industrial infrastructure it demands, the rack and fleet math, the head to head against two standalone S23 Hydros and MicroBT's competing petahash unit, the three year total cost of ownership, and a direct answer on who should buy it and who absolutely should not. Every number here was checked against Mining Now, ASIC Miner Value, OneMiners, and Bitmain's own product documentation before it went on the page.
Crossing the Petahash Barrier: Why One PH/s Per Box Is a Real Milestone
For most of mining's history, a petahash of capacity meant a wall of machines. The S19 generation that defined 2021 and 2022 ran around 100 TH/s per unit, so reaching 1 PH/s took roughly ten of them, and reaching the kind of aggregate a serious site wanted took dozens. Cabling, networking, power distribution, and failure tracking all scaled with the unit count, not with the hashrate. More boxes meant more things that could break and more hands needed to manage them.
The S23 Hyd 3U collapses that. One unit at 1.16 PH/s replaces approximately 40 S19 era machines, about 2.5 Antminer S21 XP Hydro 473TH units, or two standalone S23 Hydro 580TH units. That consolidation is the headline feature, more than the raw hashrate. Fewer Ethernet drops, fewer power whips, fewer water fittings, fewer firmware endpoints, and a lower aggregate failure rate across the cabling and connection layer where a lot of real world downtime actually originates.
This matters because mining hardware competes on two axes at once. The first is silicon efficiency, measured in joules per terahash, which determines your electricity cost. The second is deployment density, measured in hashrate per rack unit and per cooling kilowatt, which determines your infrastructure cost. The S23 Hyd 3U is the first machine that pushes hard on both at the same time. It matches Bitmain's efficiency floor and packs it into a standardized rack format built for data center deployment.
There is a quiet supply context behind why density suddenly matters so much in 2026. A wave of public miners spent the past year pivoting capacity toward AI and high performance computing, and a meaningful chunk of older Bitcoin hashrate came offline as a result. That left surviving operators competing for rack space, power contracts, and cooling capacity in facilities that were never infinite to begin with. When the constraint shifts from how many machines you can afford to how many you can physically power and cool in the space you have, a unit that delivers a full petahash per slot stops being a luxury and starts being the rational default. Density is no longer a bragging right. It is a response to a real bottleneck.
Verified Specifications, June 1, 2026
Here are the numbers, cross checked across three independent sources plus Bitmain's documentation. Where sources disagreed, that is flagged in the notes.
At 1,160 TH/s (1.16 PH/s) of SHA-256 hashrate, 11,020 watts of draw, and 9.5 joules per terahash of efficiency, the S23 Hyd 3U sits at the top of Bitmain's current lineup. Those headline numbers get repeated across every spec sheet and distributor page. What almost nobody explains is the part that actually decides whether you should buy it: the density premium. The 3U and a pair of standalone S23 Hydro units deliver nearly identical aggregate hashrate at identical efficiency. So the real question is not whether this machine is powerful. It is whether you pay extra per terahash for management consolidation, and at what fleet size that premium starts paying you back.
This review works through the verified specs as of June 1, 2026, the real profitability at four electricity rates, the strictly industrial infrastructure it demands, the rack and fleet math, the head to head against two standalone S23 Hydros and MicroBT's competing petahash unit, the three year total cost of ownership, and a direct answer on who should buy it and who absolutely should not. Every number here was checked against Mining Now, ASIC Miner Value, OneMiners, and Bitmain's own product documentation before it went on the page.
Crossing the Petahash Barrier: Why One PH/s Per Box Is a Real Milestone
For most of mining's history, a petahash of capacity meant a wall of machines. The S19 generation that defined 2021 and 2022 ran around 100 TH/s per unit, so reaching 1 PH/s took roughly ten of them, and reaching the kind of aggregate a serious site wanted took dozens. Cabling, networking, power distribution, and failure tracking all scaled with the unit count, not with the hashrate. More boxes meant more things that could break and more hands needed to manage them.
The S23 Hyd 3U collapses that. One unit at 1.16 PH/s replaces approximately 40 S19 era machines, about 2.5 Antminer S21 XP Hydro 473TH units, or two standalone S23 Hydro 580TH units. That consolidation is the headline feature, more than the raw hashrate. Fewer Ethernet drops, fewer power whips, fewer water fittings, fewer firmware endpoints, and a lower aggregate failure rate across the cabling and connection layer where a lot of real world downtime actually originates.
This matters because mining hardware competes on two axes at once. The first is silicon efficiency, measured in joules per terahash, which determines your electricity cost. The second is deployment density, measured in hashrate per rack unit and per cooling kilowatt, which determines your infrastructure cost. The S23 Hyd 3U is the first machine that pushes hard on both at the same time. It matches Bitmain's efficiency floor and packs it into a standardized rack format built for data center deployment.
There is a quiet supply context behind why density suddenly matters so much in 2026. A wave of public miners spent the past year pivoting capacity toward AI and high performance computing, and a meaningful chunk of older Bitcoin hashrate came offline as a result. That left surviving operators competing for rack space, power contracts, and cooling capacity in facilities that were never infinite to begin with. When the constraint shifts from how many machines you can afford to how many you can physically power and cool in the space you have, a unit that delivers a full petahash per slot stops being a luxury and starts being the rational default. Density is no longer a bragging right. It is a response to a real bottleneck.
Verified Specifications, June 1, 2026
Here are the numbers, cross checked across three independent sources plus Bitmain's documentation. Where sources disagreed, that is flagged in the notes.
| Specification | Value | Note |
|---|---|---|
| Hashrate | 1,160 TH/s | Bitmain tolerance ±3% (1,125-1,195 TH/s) |
| Power Consumption | 1,020 Watts | Bitmain tolerance ±5% (up to 11,571W peak) |
| Efficiency | 9.5 J/TH | Confirmed across all 3 sources |
| Cooling | Hydro (Liquid) | Requires external water loop |
| Dimensions | 900 x 486 x 132 mm | Bitmain official; fits 1U rack slot |
| Weight | 21.5 kg dry, 42 kg boxed | Confirmed; 42 kg boxed |
| Power Requirement | 380-415V 3-phase | 3-phase per Bitmain spec |
| Network Interface | Ethernet 1GbE |
One honest flag on that weight line. Mining Now lists the unit at 42 kg while MillionMiner's product page and Bitmain's manual reference 21.5 kg. The most likely explanation is that 21.5 kg is the dry chassis and 42 kg is the boxed shipping weight including packaging and the coolant manifold hardware. If your rack rails are weight rated near the margin, plan for the higher figure and verify against the shipping manifest on arrival.
Real Profitability at Four Electricity Rates
Bitcoin traded around $74,000 on June 1, 2026, down roughly 41 percent from the October 2025 all time high near $126,000, per CoinDesk pricing. Network hashrate sat near 975 to 1,020 EH/s and hashprice hovered around $35 to $37.50 per petahash per day per Hashrate Index. Against that backdrop, here is what one S23 Hyd 3U actually nets at the rates that matter.
Real Profitability at Four Electricity Rates
Bitcoin traded around $74,000 on June 1, 2026, down roughly 41 percent from the October 2025 all time high near $126,000, per CoinDesk pricing. Network hashrate sat near 975 to 1,020 EH/s and hashprice hovered around $35 to $37.50 per petahash per day per Hashrate Index. Against that backdrop, here is what one S23 Hyd 3U actually nets at the rates that matter.
The unit draws 264.48 kWh per day at 11,020 watts running flat out. That single number drives the whole table. At a hosted enterprise rate of $0.04 per kWh, electricity runs about $10.58 a day and the machine nets roughly $30.98. At MillionMiner's standard $0.07 hosting rate, electricity is $18.51 and net profit is $23.05, matching the live Mining Now figure exactly. At a typical commercial grid rate of $0.08, net falls to about $20.40. At a residential $0.12, which you could not physically use anyway given the three phase requirement, net collapses to under $10 and payback stretches past four years.
The lesson is the same one that runs through every honest hardware review: the machine is not the variable that decides your outcome. Your power rate is. A buyer at $0.04 is looking at roughly 15 month payback. The same hardware at $0.08 is looking at well over two years, and that is before you price in the difficulty drift and Bitcoin volatility that can move the whole table in a week. For the full cost-per-coin picture across hardware and rates, the cost to mine one Bitcoin breakdown runs the math across a fifteen cell matrix.
Worth noting for timing: CoinWarz projected a difficulty decrease around June 14, dropping from roughly 139T toward 124T as summer heat pulls some hashrate offline. A downward adjustment of that size lifts revenue per terahash for everyone who stays online, which would nudge every net figure in this table up by a few percent for the following epoch. That is a tailwind, not a thesis. Do not buy hardware on the strength of a single difficulty print.
The Density Premium: The Comparison Nobody Else Runs
Here is the insight that should drive your decision, and the one most reviews skip entirely. The S23 Hyd 3U and two standalone S23 Hydro 580TH units are, on paper, almost the same machine. Two Hydros give you 1,160 TH/s combined at 11,020 watts combined at 9.5 J/TH. The 3U gives you 1,160 TH/s at 11,020 watts at 9.5 J/TH. Identical hashrate. Identical power. Identical efficiency. The silicon is the same generation and the same firmware codebase.
So what are you actually paying for when you choose the 3U over two standalones? Two things, and only two things. First, rack format: the 3U mounts in a standard 19 inch cabinet and stacks cleanly, while the standalone Hydros are compact boxes that suit shelf or smaller rack placement. Second, consolidation: one management endpoint instead of two, one power whip instead of two, one set of water fittings instead of two. Halve the unit count and you halve the things that can fail at the connection layer and the things you have to touch during maintenance.
The catch is price. Across distributor channels the 3U runs roughly $28,000 to $32,500 per unit, with MillionMiner listing it at $32,379 reflecting full DDP delivery and warranty routing. Two standalone S23 Hydros typically come in below that combined, which means you are paying a per terahash premium for the rack format and the management saving. That premium is real money, and whether it is worth paying depends entirely on your fleet size.
At one or two units, the consolidation saving is trivial. Managing two endpoints versus one is not a meaningful operational burden, so the standalone route and its lower entry cost usually wins. The premium starts earning its keep when you scale. Across a 100 kW panel you are choosing between roughly 9 of the 3U units or 18 standalone Hydros for the same 10.44 PH/s. At a full megawatt, that is about 90 endpoints versus 180. At that scale, every endpoint you remove is a cable that cannot work loose, a fitting that cannot weep, a firmware push that cannot stall, and a line item your fleet management software does not have to poll. The break even is roughly where your deployment crosses from a handful of machines into the dozens, and the savings compound from there. Below that line, buy standalones. Above it, the density premium pays for itself.
Rack and Fleet Math: What the Consolidation Actually Buys You
The lesson is the same one that runs through every honest hardware review: the machine is not the variable that decides your outcome. Your power rate is. A buyer at $0.04 is looking at roughly 15 month payback. The same hardware at $0.08 is looking at well over two years, and that is before you price in the difficulty drift and Bitcoin volatility that can move the whole table in a week. For the full cost-per-coin picture across hardware and rates, the cost to mine one Bitcoin breakdown runs the math across a fifteen cell matrix.
Worth noting for timing: CoinWarz projected a difficulty decrease around June 14, dropping from roughly 139T toward 124T as summer heat pulls some hashrate offline. A downward adjustment of that size lifts revenue per terahash for everyone who stays online, which would nudge every net figure in this table up by a few percent for the following epoch. That is a tailwind, not a thesis. Do not buy hardware on the strength of a single difficulty print.
The Density Premium: The Comparison Nobody Else Runs
Here is the insight that should drive your decision, and the one most reviews skip entirely. The S23 Hyd 3U and two standalone S23 Hydro 580TH units are, on paper, almost the same machine. Two Hydros give you 1,160 TH/s combined at 11,020 watts combined at 9.5 J/TH. The 3U gives you 1,160 TH/s at 11,020 watts at 9.5 J/TH. Identical hashrate. Identical power. Identical efficiency. The silicon is the same generation and the same firmware codebase.
So what are you actually paying for when you choose the 3U over two standalones? Two things, and only two things. First, rack format: the 3U mounts in a standard 19 inch cabinet and stacks cleanly, while the standalone Hydros are compact boxes that suit shelf or smaller rack placement. Second, consolidation: one management endpoint instead of two, one power whip instead of two, one set of water fittings instead of two. Halve the unit count and you halve the things that can fail at the connection layer and the things you have to touch during maintenance.
The catch is price. Across distributor channels the 3U runs roughly $28,000 to $32,500 per unit, with MillionMiner listing it at $32,379 reflecting full DDP delivery and warranty routing. Two standalone S23 Hydros typically come in below that combined, which means you are paying a per terahash premium for the rack format and the management saving. That premium is real money, and whether it is worth paying depends entirely on your fleet size.
At one or two units, the consolidation saving is trivial. Managing two endpoints versus one is not a meaningful operational burden, so the standalone route and its lower entry cost usually wins. The premium starts earning its keep when you scale. Across a 100 kW panel you are choosing between roughly 9 of the 3U units or 18 standalone Hydros for the same 10.44 PH/s. At a full megawatt, that is about 90 endpoints versus 180. At that scale, every endpoint you remove is a cable that cannot work loose, a fitting that cannot weep, a firmware push that cannot stall, and a line item your fleet management software does not have to poll. The break even is roughly where your deployment crosses from a handful of machines into the dozens, and the savings compound from there. Below that line, buy standalones. Above it, the density premium pays for itself.
Rack and Fleet Math: What the Consolidation Actually Buys You
A full 42U rack holds about 14 S23 Hyd 3U units after accounting for manifold space and cable management, producing 16.24 PH/s per cabinet at roughly 154 kW total. Compare that against the alternatives in the same footprint. A 42U rack of S21 XP Hydro 473TH units holds about 21 machines for 9.93 PH/s, meaningfully less aggregate in the same space. The 3U wins decisively on hashrate per cabinet.
Step up to the panel level and the picture sharpens. A 100 kW industrial panel runs about 9 S23 Hyd 3U units for 10.44 PH/s aggregate. The same 100 kW of S21 XP Hydro produces only 8.04 PH/s, about 23 percent less, because the older silicon is less efficient and burns more of your power budget on the same work. The S23 Hyd 3U converts a higher share of every kilowatt into hashrate, which is the entire point of buying current generation hardware.
At megawatt scale, roughly 90 units produce around 104 PH/s. That is the kind of figure that matters to the operators MillionMiner actually serves, the ones running hosted fleets and dedicated facilities rather than a single box in a garage. For anyone weighing the broader field of current SHA-256 hardware, the best Bitcoin miners ranking for 2026 places the S23 generation against everything else worth considering by real profitability.
The Infrastructure Reality: Three-Phase, LP34, and 11 kW of Heat
This is where the S23 Hyd 3U separates the professionals from everyone else. The power requirement is not a suggestion. It needs 380 to 415V three phase service delivered through an LP34 industrial connector, the standard Bitmain specifies for this hardware. At 415V three phase the unit pulls roughly 15 amps per phase continuous and up to 25 amps per phase at peak. At 380V it pulls about 17 amps per phase continuous. Bitmain calls for dedicated 32 amp three phase breakers as a minimum for reliable continuous operation.
The LP34 connector specification alone excludes standard single phase commercial and all residential deployment. You cannot adapt your way around it with a household outlet, and you should not try. This is industrial hardware for data centers and dedicated mining facilities, full stop. Cooling is the second hard gate. Each unit dumps approximately 11 kW of thermal load, about 37,520 BTU per hour, into your facility. Handling that requires a closed loop water system with circulation pumps sized for your total fleet flow, a dry cooler or water to water heat exchanger rated for the full thermal load, inline particulate filtration, and automated flow monitoring. There is a critical interlock to know about before you energize anything: inlet water temperature must stay above 20 degrees Celsius for the miner to start, a hardware protection feature identical to the standalone S23 Hydro. Lucas learned that one the practical way, watching a freshly racked unit refuse to boot on a cold morning until the loop warmed through.
Bitmain ties your maximum facility ambient temperature to your outlet water temperature through published curves. A 35 degree outlet supports a 30 degree ambient ceiling; a 45 degree outlet supports 40 degrees. Multi unit deployments scale the cooling proportionally, usually through manifold distribution with individual isolation valves so you can pull a single unit for service without draining the whole loop. None of this is exotic for a real facility, but all of it is CapEx that has to exist before the first machine earns a satoshi.
One depth gotcha that catches buyers: the 900 mm chassis depth needs a minimum 1000 mm rack to manage cables properly, and 1200 mm is more comfortable. Standard 600 mm and 750 mm racks simply cannot take it. Check your cabinet inventory before you commit to a large order, because discovering it after delivery is an expensive surprise.
The flip side of all this infrastructure is reliability, and it is a genuine advantage. Hydro cooled hashboards run at consistent temperatures rather than cycling through the thermal swings that fatigue air cooled chips, so failure rates on hydro units run meaningfully lower over a multi year deployment. The maintenance load is real but predictable: a weekly visual check of water fittings for leaks, monthly coolant quality testing, quarterly pump and dry cooler inspection, inline filter changes when pressure differential climbs, and a full coolant replacement every 12 to 24 months. Budget roughly 3 to 5 percent of acquisition cost annually for maintenance reserves. Keep your coolant testing schedule and inlet temperature logs documented, because warranty coverage excludes damage from coolant neglect, leaks from improper fittings, and operation outside the specified environmental ranges. The paperwork is not bureaucracy. It is what protects the 365 day Bitmain warranty if a board does fail.
S23 Hyd 3U vs the MicroBT Alternative
For operators who diversify across manufacturers on purpose, and many serious fleets do precisely to avoid single vendor firmware and supply risk, MicroBT's Whatsminer M79S sits in the same petahash hydro tier at roughly 1.35 PH/s. It pushes higher raw hashrate per unit than the 3U but at a different efficiency and power profile, and it lives in MicroBT's own firmware and PSU stack rather than Bitmain's. The diversification logic that applies to the air cooled flagships in the Whatsminer versus Antminer comparison applies here too: running both vendors means a firmware bug or a warranty bottleneck on one side never takes your whole fleet down.
For most buyers choosing a single platform, the deciding factors are firmware familiarity, warranty routing in your region, and which vendor your host or facility already supports at the infrastructure level. The S23 Hyd 3U shares its firmware codebase with the rest of the S23 line and the broader Antminer fleet, so if your team already runs Antminers, the operational learning curve is effectively zero. It works with Foreman, Awesome Miner, Hive OS, and Braiins OS+ for fleet management out of the box. Three-Year Total Cost of Ownership
Hardware price is the down payment, not the cost. Here is the honest three year picture for a single self hosted unit at $0.07 per kWh, the rate band most commercial operators actually face.
Step up to the panel level and the picture sharpens. A 100 kW industrial panel runs about 9 S23 Hyd 3U units for 10.44 PH/s aggregate. The same 100 kW of S21 XP Hydro produces only 8.04 PH/s, about 23 percent less, because the older silicon is less efficient and burns more of your power budget on the same work. The S23 Hyd 3U converts a higher share of every kilowatt into hashrate, which is the entire point of buying current generation hardware.
At megawatt scale, roughly 90 units produce around 104 PH/s. That is the kind of figure that matters to the operators MillionMiner actually serves, the ones running hosted fleets and dedicated facilities rather than a single box in a garage. For anyone weighing the broader field of current SHA-256 hardware, the best Bitcoin miners ranking for 2026 places the S23 generation against everything else worth considering by real profitability.
The Infrastructure Reality: Three-Phase, LP34, and 11 kW of Heat
This is where the S23 Hyd 3U separates the professionals from everyone else. The power requirement is not a suggestion. It needs 380 to 415V three phase service delivered through an LP34 industrial connector, the standard Bitmain specifies for this hardware. At 415V three phase the unit pulls roughly 15 amps per phase continuous and up to 25 amps per phase at peak. At 380V it pulls about 17 amps per phase continuous. Bitmain calls for dedicated 32 amp three phase breakers as a minimum for reliable continuous operation.
The LP34 connector specification alone excludes standard single phase commercial and all residential deployment. You cannot adapt your way around it with a household outlet, and you should not try. This is industrial hardware for data centers and dedicated mining facilities, full stop. Cooling is the second hard gate. Each unit dumps approximately 11 kW of thermal load, about 37,520 BTU per hour, into your facility. Handling that requires a closed loop water system with circulation pumps sized for your total fleet flow, a dry cooler or water to water heat exchanger rated for the full thermal load, inline particulate filtration, and automated flow monitoring. There is a critical interlock to know about before you energize anything: inlet water temperature must stay above 20 degrees Celsius for the miner to start, a hardware protection feature identical to the standalone S23 Hydro. Lucas learned that one the practical way, watching a freshly racked unit refuse to boot on a cold morning until the loop warmed through.
Bitmain ties your maximum facility ambient temperature to your outlet water temperature through published curves. A 35 degree outlet supports a 30 degree ambient ceiling; a 45 degree outlet supports 40 degrees. Multi unit deployments scale the cooling proportionally, usually through manifold distribution with individual isolation valves so you can pull a single unit for service without draining the whole loop. None of this is exotic for a real facility, but all of it is CapEx that has to exist before the first machine earns a satoshi.
One depth gotcha that catches buyers: the 900 mm chassis depth needs a minimum 1000 mm rack to manage cables properly, and 1200 mm is more comfortable. Standard 600 mm and 750 mm racks simply cannot take it. Check your cabinet inventory before you commit to a large order, because discovering it after delivery is an expensive surprise.
The flip side of all this infrastructure is reliability, and it is a genuine advantage. Hydro cooled hashboards run at consistent temperatures rather than cycling through the thermal swings that fatigue air cooled chips, so failure rates on hydro units run meaningfully lower over a multi year deployment. The maintenance load is real but predictable: a weekly visual check of water fittings for leaks, monthly coolant quality testing, quarterly pump and dry cooler inspection, inline filter changes when pressure differential climbs, and a full coolant replacement every 12 to 24 months. Budget roughly 3 to 5 percent of acquisition cost annually for maintenance reserves. Keep your coolant testing schedule and inlet temperature logs documented, because warranty coverage excludes damage from coolant neglect, leaks from improper fittings, and operation outside the specified environmental ranges. The paperwork is not bureaucracy. It is what protects the 365 day Bitmain warranty if a board does fail.
S23 Hyd 3U vs the MicroBT Alternative
For operators who diversify across manufacturers on purpose, and many serious fleets do precisely to avoid single vendor firmware and supply risk, MicroBT's Whatsminer M79S sits in the same petahash hydro tier at roughly 1.35 PH/s. It pushes higher raw hashrate per unit than the 3U but at a different efficiency and power profile, and it lives in MicroBT's own firmware and PSU stack rather than Bitmain's. The diversification logic that applies to the air cooled flagships in the Whatsminer versus Antminer comparison applies here too: running both vendors means a firmware bug or a warranty bottleneck on one side never takes your whole fleet down.
For most buyers choosing a single platform, the deciding factors are firmware familiarity, warranty routing in your region, and which vendor your host or facility already supports at the infrastructure level. The S23 Hyd 3U shares its firmware codebase with the rest of the S23 line and the broader Antminer fleet, so if your team already runs Antminers, the operational learning curve is effectively zero. It works with Foreman, Awesome Miner, Hive OS, and Braiins OS+ for fleet management out of the box. Three-Year Total Cost of Ownership
Hardware price is the down payment, not the cost. Here is the honest three year picture for a single self hosted unit at $0.07 per kWh, the rate band most commercial operators actually face.
| Cost Component | Three-Year Total | Note |
|---|---|---|
That facility CapEx line is the one buyers underestimate most. Three phase electrical service with LP34 connectors can run $10,000 to $30,000 per facility installation. A hydro cooling loop with dry cooler infrastructure runs $10,000 to $40,000 per rack depending on capacity. Industrial rack hardware adds $2,000 to $5,000 per cabinet. Spread across a real fleet those costs amortize, but for a small deployment they can dwarf the hardware itself, which is exactly why hosting exists as a category.
This is where the hosted route changes the entire equation. Hosting eliminates the facility CapEx entirely and swaps your retail power rate for a negotiated industrial one, typically $0.04 to $0.06 per kWh. At those rates, OneMiners documented daily earnings of $68 to $95 per unit averaging $81.30 during its April to August 2025 operational data window, with ROI estimates of 8 to 13 months. The exact numbers move with Bitcoin's price, but the structural point holds: the cheaper your power and the lower your infrastructure burden, the faster the machine pays back. The full line by line of what a hosting rate actually covers is broken down in the ASIC hosting cost breakdown.
Who Should Buy It, and Who Absolutely Should Not
This is where the hosted route changes the entire equation. Hosting eliminates the facility CapEx entirely and swaps your retail power rate for a negotiated industrial one, typically $0.04 to $0.06 per kWh. At those rates, OneMiners documented daily earnings of $68 to $95 per unit averaging $81.30 during its April to August 2025 operational data window, with ROI estimates of 8 to 13 months. The exact numbers move with Bitcoin's price, but the structural point holds: the cheaper your power and the lower your infrastructure burden, the faster the machine pays back. The full line by line of what a hosting rate actually covers is broken down in the ASIC hosting cost breakdown.
Who Should Buy It, and Who Absolutely Should Not
Buy the S23 Hyd 3U if you run an industrial or hosted operation with 380 to 415V three phase power and a hydro loop already in place, or if you are placing units into a host that provides that infrastructure. Buy it if you are deploying three or more units and you value cutting your endpoint and connection count. Buy it if your power sits at or below $0.07 per kWh and you optimize for hashrate per rack and per cooling kilowatt. For a facility built around current generation hydro at scale, it is the strongest density and efficiency play Bitmain currently sells.
Consider the standalone S23 Hydro 580TH instead if you only want one or two units, where the density premium does not pay for itself and the lower per terahash entry cost wins. Same 9.5 J/TH efficiency, lower commitment, more placement flexibility.
Do not buy it for a home or residential setup, period. There is no three phase service or LP34 connector in a house, no practical way to handle 11 kW of thermal load per unit, and at residential power rates above $0.12 the margin evaporates regardless. If you want hydro cooling at home, the Antminer S21+ Hydro 395TH with universal single phase voltage is the residential ceiling. The honest framing matches the one in the home versus hosted mining breakdown: the right machine for you is the one your power and infrastructure can actually support, not the one with the biggest number on the box. And if you are still deciding whether mining makes sense for your situation at all, the five real profitability scenarios give you concrete yes or no answers.
What Could Break This Thesis
Honesty section, because every hardware decision carries risk. Bitcoin at $74,000 is down sharply from its 2025 peak, and the entire profitability table here moves with the spot price. A further drop pressures every margin in this review. Difficulty is the other lever: if hashrate recovers faster than expected after the summer dip, difficulty climbs and revenue per terahash falls. The June 14 downward adjustment is a short term tailwind, not a durable trend.
There is also the obsolescence clock. The S23 Hyd 3U is current generation today, but Bitmain and MicroBT both iterate aggressively, and a more efficient unit will arrive. Hardware bought near a price peak and operated at marginal power rates is the classic way operators lose money in this business. The machine is excellent. Whether it is excellent for you depends on power, infrastructure, fleet size, and your read on Bitcoin, none of which a spec sheet can answer for you.
Frequently Asked Questions
What hashrate and efficiency does the S23 Hyd 3U deliver? It delivers 1.16 PH/s (1,160 TH/s) of SHA-256 hashrate at 11,020 watts and 9.5 J/TH efficiency. Bitmain's plus or minus 3 percent tolerance means real output ranges 1,125 to 1,195 TH/s, and the plus or minus 5 percent power tolerance means wall draw can reach 11,571 watts at peak. It is the first production Bitcoin miner to cross one petahash per unit.
Is the S23 Hyd 3U profitable in 2026?
Yes, at the right power rate. On June 1, 2026, at Bitcoin near $74,000, one unit nets roughly $23.05 per day at $0.07 per kWh and about $30.98 at a hosted $0.04 rate, with gross revenue around $41.56 per day. At residential rates above $0.12 the margin nearly disappears. Profitability tracks your electricity cost more than any other factor.
How is the 3U different from the standalone S23 Hydro 580TH?
Form factor and density, not performance. Two standalone Hydros equal one 3U in hashrate, power, and efficiency. The 3U mounts in a standard 19 inch rack and consolidates two machines into one endpoint, one power whip, and one set of fittings. The standalone costs less per terahash. Choose the 3U at fleet scale where consolidation pays; choose standalones for one or two units.
What power infrastructure does the S23 Hyd 3U need?
380 to 415V three phase service through an LP34 industrial connector, with dedicated 32 amp three phase breakers minimum. At 415V it draws about 15 amps per phase continuous and up to 25 at peak. The LP34 connector rules out single phase commercial and all residential deployment. This is strictly industrial hardware.
Can I run the S23 Hyd 3U at home?
No. Home electrical service cannot provide 380 to 415V three phase power or an LP34 connector, and a house cannot practically handle 11 kW of thermal load per unit. At residential rates the economics fail anyway. For home hydro, the Antminer S21+ Hydro 395TH with universal single phase voltage is the residential ceiling.
How many older miners does one S23 Hyd 3U replace?
Roughly 40 Antminer S19 era units at 100 TH/s each, about 2.5 S21 XP Hydro 473TH units, or two standalone S23 Hydro 580TH units. The consolidation cuts Ethernet drops, PSU cables, water fittings, and firmware endpoints, lowering the aggregate failure rate at the connection layer where much real world downtime starts.
What is the fleet and rack math?
A 42U rack holds about 14 units producing 16.24 PH/s at roughly 154 kW. A 100 kW panel runs about 9 units for 10.44 PH/s, versus 8.04 PH/s for the same power in S21 XP Hydro hardware, about 23 percent more aggregate from the newer silicon. At a megawatt, roughly 90 units produce around 104 PH/s.
What cooling does the S23 Hyd 3U require?
A closed loop water system rated for about 11 kW of thermal load per unit (37,520 BTU per hour), with circulation pumps, a dry cooler or heat exchanger, inline filtration, and flow monitoring. Inlet water must stay above 20 degrees Celsius or the unit will not start, a hardware interlock. Maximum facility ambient temperature scales with outlet water temperature per Bitmain's curves.
How much does the S23 Hyd 3U cost?
Across distributor channels it runs roughly $28,000 to $32,500 per unit. MillionMiner lists it at $32,379 with free DDP worldwide delivery, customs clearance included, and 365 day Bitmain warranty routing, which accounts for the premium end of that range. Bulk orders of ten or more units qualify for custom pricing.
What is the realistic ROI on the S23 Hyd 3U?
At hosted industrial rates of $0.04 per kWh, payback runs roughly 8 to 15 months per OneMiners data and current conditions. At $0.07 to $0.08 commercial rates it stretches to 20 to 27 months or more, depending heavily on Bitcoin's price and network difficulty. Lower power and lower infrastructure burden through hosting shorten payback the most.
The Bottom Line
Back in Sheridan, Lucas did not buy the S23 Hyd 3U because it had the biggest number on the box. He bought it because his facility already had the three phase service, the hydro loop, and the rack depth to support it, and because at his fleet size every endpoint he removed was one less thing to chase at three in the morning. The forty S19 units he pulled were not broken. They were simply costing him more in power, space, and attention than a single current generation slot. That is the entire case for this machine, and it is also the entire case against it for anyone whose situation does not look like his.
The S23 Hyd 3U is the most capable single Bitcoin miner Bitmain currently ships, and on a real industrial or hosted footprint it is hard to beat on density and efficiency together. It is also completely wrong for a home, a single phase shop, or a one unit hobbyist, and the price premium over standalone Hydros only earns out once your fleet grows into the dozens. Match the machine to your power and your infrastructure, run your own numbers at your own rate, and the decision answers itself.
Consider the standalone S23 Hydro 580TH instead if you only want one or two units, where the density premium does not pay for itself and the lower per terahash entry cost wins. Same 9.5 J/TH efficiency, lower commitment, more placement flexibility.
Do not buy it for a home or residential setup, period. There is no three phase service or LP34 connector in a house, no practical way to handle 11 kW of thermal load per unit, and at residential power rates above $0.12 the margin evaporates regardless. If you want hydro cooling at home, the Antminer S21+ Hydro 395TH with universal single phase voltage is the residential ceiling. The honest framing matches the one in the home versus hosted mining breakdown: the right machine for you is the one your power and infrastructure can actually support, not the one with the biggest number on the box. And if you are still deciding whether mining makes sense for your situation at all, the five real profitability scenarios give you concrete yes or no answers.
What Could Break This Thesis
Honesty section, because every hardware decision carries risk. Bitcoin at $74,000 is down sharply from its 2025 peak, and the entire profitability table here moves with the spot price. A further drop pressures every margin in this review. Difficulty is the other lever: if hashrate recovers faster than expected after the summer dip, difficulty climbs and revenue per terahash falls. The June 14 downward adjustment is a short term tailwind, not a durable trend.
There is also the obsolescence clock. The S23 Hyd 3U is current generation today, but Bitmain and MicroBT both iterate aggressively, and a more efficient unit will arrive. Hardware bought near a price peak and operated at marginal power rates is the classic way operators lose money in this business. The machine is excellent. Whether it is excellent for you depends on power, infrastructure, fleet size, and your read on Bitcoin, none of which a spec sheet can answer for you.
Frequently Asked Questions
What hashrate and efficiency does the S23 Hyd 3U deliver? It delivers 1.16 PH/s (1,160 TH/s) of SHA-256 hashrate at 11,020 watts and 9.5 J/TH efficiency. Bitmain's plus or minus 3 percent tolerance means real output ranges 1,125 to 1,195 TH/s, and the plus or minus 5 percent power tolerance means wall draw can reach 11,571 watts at peak. It is the first production Bitcoin miner to cross one petahash per unit.
Is the S23 Hyd 3U profitable in 2026?
Yes, at the right power rate. On June 1, 2026, at Bitcoin near $74,000, one unit nets roughly $23.05 per day at $0.07 per kWh and about $30.98 at a hosted $0.04 rate, with gross revenue around $41.56 per day. At residential rates above $0.12 the margin nearly disappears. Profitability tracks your electricity cost more than any other factor.
How is the 3U different from the standalone S23 Hydro 580TH?
Form factor and density, not performance. Two standalone Hydros equal one 3U in hashrate, power, and efficiency. The 3U mounts in a standard 19 inch rack and consolidates two machines into one endpoint, one power whip, and one set of fittings. The standalone costs less per terahash. Choose the 3U at fleet scale where consolidation pays; choose standalones for one or two units.
What power infrastructure does the S23 Hyd 3U need?
380 to 415V three phase service through an LP34 industrial connector, with dedicated 32 amp three phase breakers minimum. At 415V it draws about 15 amps per phase continuous and up to 25 at peak. The LP34 connector rules out single phase commercial and all residential deployment. This is strictly industrial hardware.
Can I run the S23 Hyd 3U at home?
No. Home electrical service cannot provide 380 to 415V three phase power or an LP34 connector, and a house cannot practically handle 11 kW of thermal load per unit. At residential rates the economics fail anyway. For home hydro, the Antminer S21+ Hydro 395TH with universal single phase voltage is the residential ceiling.
How many older miners does one S23 Hyd 3U replace?
Roughly 40 Antminer S19 era units at 100 TH/s each, about 2.5 S21 XP Hydro 473TH units, or two standalone S23 Hydro 580TH units. The consolidation cuts Ethernet drops, PSU cables, water fittings, and firmware endpoints, lowering the aggregate failure rate at the connection layer where much real world downtime starts.
What is the fleet and rack math?
A 42U rack holds about 14 units producing 16.24 PH/s at roughly 154 kW. A 100 kW panel runs about 9 units for 10.44 PH/s, versus 8.04 PH/s for the same power in S21 XP Hydro hardware, about 23 percent more aggregate from the newer silicon. At a megawatt, roughly 90 units produce around 104 PH/s.
What cooling does the S23 Hyd 3U require?
A closed loop water system rated for about 11 kW of thermal load per unit (37,520 BTU per hour), with circulation pumps, a dry cooler or heat exchanger, inline filtration, and flow monitoring. Inlet water must stay above 20 degrees Celsius or the unit will not start, a hardware interlock. Maximum facility ambient temperature scales with outlet water temperature per Bitmain's curves.
How much does the S23 Hyd 3U cost?
Across distributor channels it runs roughly $28,000 to $32,500 per unit. MillionMiner lists it at $32,379 with free DDP worldwide delivery, customs clearance included, and 365 day Bitmain warranty routing, which accounts for the premium end of that range. Bulk orders of ten or more units qualify for custom pricing.
What is the realistic ROI on the S23 Hyd 3U?
At hosted industrial rates of $0.04 per kWh, payback runs roughly 8 to 15 months per OneMiners data and current conditions. At $0.07 to $0.08 commercial rates it stretches to 20 to 27 months or more, depending heavily on Bitcoin's price and network difficulty. Lower power and lower infrastructure burden through hosting shorten payback the most.
The Bottom Line
Back in Sheridan, Lucas did not buy the S23 Hyd 3U because it had the biggest number on the box. He bought it because his facility already had the three phase service, the hydro loop, and the rack depth to support it, and because at his fleet size every endpoint he removed was one less thing to chase at three in the morning. The forty S19 units he pulled were not broken. They were simply costing him more in power, space, and attention than a single current generation slot. That is the entire case for this machine, and it is also the entire case against it for anyone whose situation does not look like his.
The S23 Hyd 3U is the most capable single Bitcoin miner Bitmain currently ships, and on a real industrial or hosted footprint it is hard to beat on density and efficiency together. It is also completely wrong for a home, a single phase shop, or a one unit hobbyist, and the price premium over standalone Hydros only earns out once your fleet grows into the dozens. Match the machine to your power and your infrastructure, run your own numbers at your own rate, and the decision answers itself.
Ready to Start Mining?
Free worldwide DDP shipping. Professional hosting from $0.055/kWh.
