Inside a GoMining farm in Texas

Almost everything written about GoMining is a screen: an app, a dashboard, a number going up. What you almost never see is the iron. Which is why this video is useful even though it contains no profitability figure at all: it shows the part of the business you cannot fake with a screenshot.

It is a tour of a second Texas building, filmed by the VoskCoin channel, with two people from the operation answering on camera: the executive hosting the visit, and one of the technicians who works the site every day. What follows is what they say, put in order, and what can be derived from it with this site's calculator.

The first thing you see is a building switched off

They arrive on site and there is no noise. Nothing is mining. And that — which looks like the worst possible opening shot for a tour — turns out to be the most valuable explanation in the whole video.

It is called curtailment, and it is not a fault: it is the deal. The building shuts down during the hours when the grid is under most strain. The specific mechanism they name is 4CP — the four coincident peaks of demand on the Texas grid: if you manage to be offline at exactly those moments, your transmission fees are waived. The better you predict them, the better the bill.

The way they sum it up is that the farm works like a battery for the grid. And along the way they take apart the idea that a miner is just somebody who plugs in: the comparison they draw is with AI data centres, which arrive demanding hundreds of firm megawatts and never switch off, and which they blame for the hostile mood now aimed at data centres in general. The other half of the argument is that they go after spare capacity at substations that already exist — infrastructure the grid has already paid for and was under-using.

For anyone looking at this with a calculator in hand, there is the answer to a question that comes up a lot: why electricity is $0.05–0.07/kWh and not what anyone pays at home. It is not a volume discount. It is that they agree to switch off when the grid needs it, and that is what the rate costs. It is the constant the methodology flags as verified against official documentation, and this video is the first time you get to see where it physically comes from.

The scale, in their own figures

These are the numbers said on camera. Nobody has audited them — this site included — and they should be read for what they are: what the company says about itself, with the advantage that it says it standing inside the building with the camera rolling.

Figures stated in the video by the people running the operation, in July 2026. They qualify the pipeline number themselves on camera: those are sites under evaluation, and most of them will most likely never be acquired.

The 95,000 is worth pausing on. It is not the machine count of this building: it is what the company's monitoring team handles across the whole operation, with in-house software, VPN access and every machine tracked by serial number. That is the difference between "we have some buildings" and "we have an operation".

The question the video asks and never answers

It opens with three questions: how many machines are on site, what that is in hashrate, and what the power bill looks like. The first one gets answered — 6,800. The other two hang in the air for seventeen minutes.

They can be derived, and the derivation is more honest done from the megawatts than from the machine count: 25 MW is a hard number, and whichever miner model sits inside does not change what the building draws. What the model changes is how much hashrate you get out of those same watts — which is exactly what W/TH measures.

All 25 MW dedicated to mining, with electricity at $0.05/kWh and at the per-TH reward this site computes from live network data. It is a ceiling: some of those megawatts go to cooling, and the building is offline during peaks, so real hashrate is lower. The last column is the electricity price at which that building would stop making money mining.

On the 27 July 2026 snapshot — network at 867 EH/s, bitcoin at $64,943 — a 25 MW building at 15 W/TH lands around 1.67 EH/s, roughly 0.19% of all the bitcoin being mined on Earth. And it produces about $53,800 a day in bitcoin.

The bill, by contrast, does not depend on efficiency: 25 MW running 24 hours is 600,000 kWh a day, and at $0.05/kWh that is $30,000 of electricity daily. Fixed. Whatever is plugged in.

And this is where the video and the calculator shake hands. At one point in the tour there is a joke about a site offered at 10.5 cents per kilowatt-hour. At that price the same building would pay $63,000 a day in power to mine $53,800. It would lose nearly ten thousand dollars a day without a single thing breaking.

That is why they switch off at peaks. It is not corporate social responsibility: it is that the entire business lives between 5 and 9 cents, and anything pushing the rate above that kills it.

It is exactly your problem, multiplied

Here is what makes this video interesting for someone who only holds 100 TH in an app.

In how much bitcoin 1 TH mines per day it works out that an efficient terahash clears about three dollars a year, and that two identical miners three watts apart end up 34% apart in profit. The 25 MW building is that same sum with more zeros: same reward per TH, same split between income and bill, same thin margin.

With one difference worth seeing, because it inverts the intuition:

• For you, efficiency changes what you pay. Your TH are fixed — you bought them — and a worse machine raises your power bill against the same income.
• For the building, efficiency changes what it earns. Its megawatts are fixed — that is what the substation gave it — and a worse machine lowers its hashrate against the same bill.

Same fraction, seen from both sides. And it explains why there are whole rows of XP machines in the video: not for fashion, but because on a site where megawatts are capped, W/TH is literally revenue.

What breaks, and who fixes it

The dullest part of the video is the one that says most about the real risk of buying hosted hashrate. The technician describes the day: get in around seven, open the laptop, look at the issue list and start. What they touch daily is power cycling, fans and power supplies.

And there is an explicit limit: they do not open hashboards or motherboards. Not because they cannot, but because most of the machines are under warranty and repairing them in-house would void it. Those go to the warranty centre.

Translated into what matters to a buyer: you do not pay for the failure and you do not do the work, but nobody pays for the time a machine spends down. And that is exactly one of the places the gap this site has spent months measuring comes from.

The calculator's model applies a pool factor of 0.952: miners do not receive the network's theoretical reward, but somewhat less. That factor has been measured twice against real payouts, giving 0.952 and 0.948. What has never been possible to break down is what that ~5% is made of: pool fee, platform overhead, luck… and downtime. After watching the video there are two fairly obvious new candidates for that list: peak curtailment, and machines waiting on parts.

This is a hypothesis, not a finding. GoMining does not publish the breakdown and the video does not give it either. What is measured is the aggregate result: the reward that arrives is 4.8% below theoretical, and the calculator applies that. The detail, with its limits, is in the methodology.

Water, the problem nobody sees coming

Cooling is evaporative: water over panels, with outside air pulled through them. What that needs is water, and there is no well on that site.

The reason is beautifully Texan: they drilled and hit oil. On top of that, a well permit can cap you at 25,000 gallons a day when a large site might want 100,000 on a peak day. The stopgap was tanker trucks filling a tank — expensive and wasteful, as they admit themselves — and the permanent fix is piping water from a reservoir a couple of kilometres away.

Then there is the hot aisle, sealed with polycarbonate sheeting so the exhaust air cannot loop back round to the intake. The technician says that not even rain gets in: the fan pressure keeps it out. It is a small detail and it is exactly the kind of thing you only learn by operating.

None of this — the water, the land, the substation, the permit — appears in any mining calculator on Earth, ours included. But it is what decides whether a building exists at all. When they say in the video that they have 2 to 3 GW under evaluation and that most of them will probably never be bought, this is what they are talking about.

What this video proves and what it does not

This part is the important one, because a facility tour is persuasive content by nature and it pays to be clear about what you walk away with.

What it does prove:

• That the iron exists and is at industrial scale. 20 and 25 MW buildings, thousands of machines per row, standardised racks and PDUs across sites. That is not staged.
• That there is an operation behind it, not just an app. In-house monitoring with serial-level inventory, repair teams, a warranty policy, technicians on site.
• That the electrical terms are what they claim to be. Curtailment and 4CP coherently explain the $0.05–0.07/kWh range the platform documents. It fits.
• That they understand the cycle. Talking about preparing for the downturn while everyone is celebrating is, in this sector, close to a rarity.

What it does not prove, and should not be allowed to look like it does:

• It says nothing about your returns. Not one number in the video enters the sum of what a digital miner earns. Returns are set by the bitcoin price, difficulty, your W/TH and your discounts, and none of those four improve because the building is big.
• No figure is verified. The 25 MW, the 6,800 miners and the 95,000 machines are statements. Credible, internally consistent, and unaudited.
• It is not independent journalism, and you do not have to infer that. The video carries YouTube's paid promotion label: the channel declares having received money or gifts to make it. That fits what you see inside, where it is given for free the tour others are charged for. It does not invalidate it — it shows real things, and the label is the opposite of hiding it — but this is sponsored content, not an audit.
• And a solid company can still be a bad deal. They are two different questions: "does this exist and work?" and "is it worth it for me?". This video answers the first one very well. The second is what the calculator does, and for a fair number of configurations its answer is no.

In short

• The building was off on purpose. 4CP curtailment: they shut down at grid peaks in exchange for waived transmission fees. That is the real origin of $0.05–0.07/kWh.
• 25 MW and 6,800 miners on this site; around 3,000 per row; 95,000 machines monitored company-wide and roughly 254 people.
• Those 25 MW are on the order of 1.7 EH/s at 15 W/TH: about 0.19% of the network, some $53,800 mined daily against $30,000 of power.
• The bill does not depend on efficiency; the hashrate does. For the building, W/TH is revenue; for you it is cost. Same fraction, opposite sides.
• The business lives between 5 and 9 cents per kWh. At the 10.5 joked about in the video, that same building would lose nearly $10,000 a day.
• And none of it changes your sum. It lends credibility to the infrastructure, not profitability to your miner. Two different questions, worth not confusing.