How AI Smart-Switch mining works

Why hashpower should be allowed to move

ShiftHash is built around one idea: instead of pointing your rented hashpower at a single coin and hoping that coin stays the most rewarding choice, the system keeps checking the market and moves your hashpower when another coin becomes clearly more rewarding. That is what AI Smart-Switch cloud mining means in practice. You buy a plan, that plan gives you a fixed amount of hashpower, and the optimizer decides, again and again through the day, which proof-of-work coin that hashpower should mine. The goal is not to chase every small price movement. The goal is to catch the larger, meaningful differences in estimated yield and ignore the noise in between.

Single-coin mining leaves yield on the table because coins do not move together. Prices rise and fall at different times, mining difficulty changes as miners join or leave a network, and block rewards are worth different amounts from one week to the next. A coin that was the best choice on Monday can be an average choice by Thursday, and a coin that looked weak last month can become the strongest choice after a price move or a difficulty drop. If your hashpower is locked to one coin, you keep earning at that coin's rate whether it is leading or lagging. If your hashpower can move, it can follow the better estimated rate instead. That is the whole case for Smart-Switch: the market keeps changing, so the allocation should be allowed to change with it, within rules that keep the switching sensible. ShiftHash evaluates twelve proof-of-work coins: BTC, BCH, LTC, DOGE, KAS, ETC, RVN, ZEC, DASH, ERG, FLUX and DGB. ETH is not in the list, and that is deliberate. ETH is proof-of-stake, which means it is not mineable with hashpower in the way these coins are, so including it in a mining comparison would be meaningless. The twelve coins on the list are the ones the optimizer can actually compare on mining yield. Two of them need a special note. DOGE is merged-mined with LTC, which means DOGE revenue folds into the LTC score rather than being counted as a separate, competing option. Counting it twice would overstate what the same hashpower could earn, so the system counts it once, inside LTC. FLUX is currently excluded because no reliable free difficulty and reward feed was found for it. It stays on the list of coins the product knows about, but it is not scored for switching until a trustworthy feed exists. Excluded coins are shown on the Optimizer screen with the reason, so you can see what is left out and why, instead of wondering whether the system simply forgot it.

Every 15 minutes: how an evaluation works

An evaluation happens every fifteen minutes. That interval matters. Fifteen minutes is frequent enough to catch real changes during the day, but not so frequent that the system reacts to every passing tick. Each evaluation follows the same steps. First, the optimizer pulls the live price for each coin from CoinGecko. Second, it pulls live difficulty data from WhatToMine and mempool.space. Price tells the system what a mined coin is worth right now. Difficulty tells the system how hard it is to mine that coin right now, which is another way of saying how much of the coin a given amount of hashpower is likely to produce. A high price means little if the difficulty is so high that you mine very little of the coin, and a low difficulty means little if the coin is worth very little. The two have to be combined.

The score: estimated net yield per watt per day

The combination is the score: estimated net yield in USD per watt per day. That unit is the heart of the whole system, so it is worth understanding. Per watt per day puts every coin on the same footing. It answers the question of how many dollars of net value one watt of mining power is estimated to produce in a day on that coin. The optimizer does not compare raw coin prices, and it does not compare raw amounts of coins mined. It compares this single, normalised yield figure. A coin that produces a large number of cheap coins and a coin that produces a small number of expensive coins can then be compared fairly, because both are reduced to the same question: what is the estimated dollar yield per watt per day.

The word net in that score matters too. The score is calculated after a one percent pool fee is taken off. The pool fee is the cost of mining through a pool rather than alone, and the optimizer subtracts it before comparing coins, so the number you see is the estimated yield after that cost. That keeps the comparison honest. A coin that looks slightly better before fees but worse after them should not win, and with the fee applied inside the score, it will not.

The 5% rule: the noise filter

Once every coin has a score, the optimizer compares the best score with the score of the coin your hashpower is currently mining. This is where the five percent rule comes in. The system switches hashpower only when another coin beats the current pick by at least five percent of estimated net advantage. If the best alternative is three percent better, or one percent better, or exactly equal, the system stays put. That is not hesitation and it is not a delay. It is the design. Small differences are often noise: a price that moved a little, a difficulty reading that will move back, a gap that would disappear by the time a switch had finished. Switching on a two percent gap would cost time and disruption for a gain that might not survive the next evaluation fifteen minutes later. The five percent threshold is a noise filter. It says that a switch should only happen when the advantage is large enough to be worth acting on, and clear enough that it is probably real rather than a flicker.

Daily switch budgets

There is a second guard as well: daily switch limits. Each plan tier has a maximum number of successful switches per day. The twenty dollar tier allows one switch per day, the one hundred dollar tier allows five, the three hundred dollar tier allows fifteen, the five hundred dollar tier allows thirty, and the one thousand dollar tier allows sixty. Only successful switches count toward the limit, and the limit resets at midnight UTC. The limits stack across active contracts, so if you hold more than one active contract, your total daily budget is the sum of their limits. These limits exist for the same reason as the threshold: to keep the system from switching too often. Even a genuine advantage is worth less if the system switches back and forth all day, so the budget forces the optimizer to spend its switches where they matter most. If the budget for the day is used up, the optimizer will wait until the reset, even if a new advantage appears. That is expected behaviour, and the Optimizer screen shows how many switches have been used and how many remain, so it is never a mystery.

Locking the optimizer to one coin

You can also lock the optimizer to a single coin. Locking pauses auto-switching entirely and keeps your hashpower on the coin you chose. That is useful if you have a strong preference for holding a particular coin, or if you simply want your mining to stay predictable for a while. While locked, the fifteen-minute evaluations still produce information, but they do not move your hashpower.

Plans, ROI and hashrate

The plans themselves are straightforward. The tiers are twenty, one hundred, three hundred, five hundred and one thousand dollars per month, with terms of one, three, six or twelve months. Longer terms cost less per month: ten percent off for three months, twenty percent off for six months, and thirty percent off for twelve months. The return on investment figures for the tiers are one hundred and eighty percent for the twenty dollar tier, two hundred and ten percent for the one hundred dollar tier, two hundred and forty percent for the three hundred dollar tier, two hundred and seventy percent for the five hundred dollar tier, and three hundred percent for the one thousand dollar tier. Hashrate rises with the tier: one hundred and sixty terahashes per second for the twenty dollar tier, eight hundred for the one hundred dollar tier, two thousand four hundred for the three hundred dollar tier, four thousand for the five hundred dollar tier, and eight thousand for the one thousand dollar tier. Plans are stackable, so you can hold more than one at a time, and as noted above, their switch limits stack too. Every figure in this paragraph is an estimate, never guaranteed profit. Prices move, difficulty moves, and the return on a mining plan depends on conditions that no one can promise in advance.

Terms of 1, 3, 6 or 12 months. Longer terms cost less per month: 10% off for 3 months, 20% off for 6 months, 30% off for 12 months. Every figure is an estimate, never guaranteed profit.

Proof it works: the first live switch

The system has already shown that it will switch when the data justifies it. On November 5, 2026, the engine made its first live auto-switch, moving allocation from BTC to ZEC at an eight hundred and forty-four point six zero percent estimated advantage. At that moment ZEC was scoring around zero point zero one eight one seven dollars per watt per day, while BTC was scoring around zero point zero zero one nine two dollars per watt per day. That is not a small edge. It is the kind of gap the five percent threshold exists to catch: large, clear, and worth acting on. The switch happened because the numbers said so, not because someone guessed that ZEC would do well. It is one event, and one event does not prove that every day will look like that, but it does prove the mechanism works end to end: live prices in, live difficulty in, scores compared, threshold passed, hashpower moved.

Payouts and the Optimizer screen

What you earn is paid out in BTC or USDT. Your mining earnings accrue as a mining balance, and from there you move them with Collect or Convert. Collect moves the entire mining balance to your withdrawable balance, ready for payout. Convert moves the entire mining balance to your account balance, where it can be spent inside ShiftHash. Both moves carry no fee, and both move the whole balance, never part of it. The BTC withdrawal minimum is zero point zero zero zero three BTC, with a network fee of zero point zero zero zero five BTC, and withdrawals are reviewed before they are sent. Payout in USDT is also available. The choice between BTC and USDT is a choice about which asset you want to receive, not a choice that changes how the optimizer scores coins.

The Optimizer screen is where all of this becomes visible. It shows the time of the last evaluation, so you can see how fresh the data is. It shows the best coin and its advantage against the five percent threshold, so you can see how close the system is to switching. It shows the per-coin scored yields, so you can compare the coins yourself instead of taking the verdict on trust. It shows excluded coins with their reasons, so FLUX and any other exclusion is explained in place. It shows switches used against the budget for today, so you know how much switching room remains. And it shows a plain-language why-no-switch verdict, which explains in ordinary words why the system is staying put when it is staying put. If you read that screen regularly, you will understand your mining far better than a single earnings number could ever tell you. You will see the market as the optimizer sees it: a set of competing yields, a threshold, a budget, and a decision. Prices can move sharply in either direction between evaluations, difficulty can change as miners move between networks, and a coin that leads today can lag tomorrow. A feed can also go stale or go missing, and when that happens the system skips the evaluation rather than switching on bad data. Skipping is the safe choice: a switch made on stale prices or a missing difficulty reading could move your hashpower for the wrong reason, and a missed evaluation costs far less than a wrong switch. The fifteen-minute rhythm, the five percent threshold, the daily budgets, the fee inside the score, and the skip-on-stale-feed rule are all parts of the same approach: move when the evidence is strong, and otherwise wait. Smart-Switch is not a promise that switching always wins. It is a disciplined way of letting the numbers decide, frequently, transparently, and only when the advantage is big enough to matter.