Difficulty Adjusted — July 2026
Issue #3 · The June close: my first month underwater, and the number that actually mattered.
The Adjustment
Difficulty sits near 127 trillion as I write this, down 5 percent at the adjustment that landed today, with the next recalibration due toward the end of July. That makes three sharp moves in four weeks: down 10 percent in mid-June, most of it clawed back two weeks later, now down 5 again. The network read a violent month and moved the target every time, without flinching in any direction. This issue is about the month my own numbers asked for the same treatment. The edge I quote every month inverted, and I had to sit with what that does and doesn’t change.
June was the month the operation stopped flattering me. The fleet finished nine dollars underwater, my first negative month since the reframe, and the below-spot edge I quote in every issue inverted: it cost me slightly more to mine a coin than to buy one. Nothing on the racks changed; bitcoin fell 19 percent to a 21-month low, dragged by record ETF outflows and a Fed that turned hawkish, and at these prices something like a fifth of miners are underwater with me. That’s a liquidity verdict, not a Bitcoin verdict. So this issue is about reading a bad print without flinching, and what the framework actually does when the edge goes away for a month.
Fleet pulse
Three S21-class rigs, three hosts, ~667 TH/s combined, unchanged. June produced 0.00867 bitcoin on $548 of gross revenue against $557 of hosting, putting the fleet net at −$9 for the month, about −$0.30/day. That’s the first net-negative month since I reframed this operation in the spring, and I’m not going to dress it up.
I will say plainly what it was and wasn’t. The rigs did the same work they did in May. Bitcoin averaged around $63K in June against $78K in May, down roughly 19 percent, and the fleet mined about 3 percent fewer coins into that weaker price. Nothing operational broke. The machines hashed, the hosts invoiced, the pools paid. The month was a price event that showed up on my P&L wearing a mining costume. What it actually changed sits in the next section.
The strategy layer
Last issue was about not letting a good month talk me into expanding. This one is the mirror image: not letting a bad month talk me into quitting early.
The number that matters isn’t the −$9. It’s this one: my hosting-only cost basis came in at $64,266 per bitcoin mined in June, against a June average spot around $63,000. For the first time since I started publishing this, the operation mined bitcoin slightly above market. The below-spot edge I’ve quoted every month inverted. In May the same math said $61K basis against $78K spot. Same fleet, five weeks later, opposite sign. That’s what a 19 percent price move does to a spread that was never as wide as a good month makes it look.
So what does the framework actually do with that? Two things, both mechanical. First, the marginal rig printed its worst month yet, and that starts my kill clock: the rule is two consecutive net-negative closes and the rig comes off. One month starts the clock; it doesn’t fire it. The July close is the decision point, and I’ve already committed to letting the rule decide rather than my mood. Second, one of my “negative” rigs turned out to be negative entirely because of a 3.5 percent card-processing fee on the hosting invoice. Except even that isn’t the full story: the card pays 1.5 percent back in bitcoin, so the real drag on that rail is about 2 percent, and once I count the sats, the rig actually cleared the month. That’s a few dollars in either direction, but it’s the whole ethos in miniature: read the invoice line by line, then net out the payment rail on top of it, before you conclude anything about your operation. Sometimes the problem is neither the machine nor the fee; it’s the column you forgot to count.
What carried the month was neither of those. Hosting spend is deductible, and at a 32 percent marginal rate, $557 of hosting throws off roughly $178 of tax shield. After tax, the fleet’s worst pre-tax month of the year still nets out meaningfully positive. The spread lapsed in June. The structure didn’t.
Extended strategy
The honest version: I spent a week wanting to front-run my own rule.
When the marginal rig printed −$36 into a sub-$60K tape, the case for retiring it immediately wrote itself. Price is in my red zone, the spread inverted, and waiting a month for a second data point felt like formality. Why pay another hosting invoice to confirm what I already know?
Because I don’t already know it. That’s the whole point of the rule. June’s loss was a price event, and price events reverse without asking permission; retiring a rig doesn’t. Pull it now and I give up a machine on 0 percent financing, crystallize the exit at the bottom of the month’s range, and take a decision that took thirty seconds to feel right and would take months to undo. The two-close rule exists precisely because one month, in either direction, is a single observation, and I already wrote an entire issue about not acting on single observations when they pointed up.
The other temptation was quieter: to lead this issue with the after-tax number and bury the inversion. I didn’t, and the fact that I wanted to is worth admitting.
Tax & leverage corner: DRIP on the bitcoin preferreds, where the dividend is a return of capital
Part of my treasury sits in one of the bitcoin-treasury perpetual preferreds (Strategy’s STRC and Strive’s SATA are the two most visible), with the DRIP switched on so every monthly payout buys fractional shares automatically.
Here’s the part most operators get wrong: these aren’t ordinary dividends. Because the issuers have no accumulated earnings and profits and don’t expect any, the distributions are classified as return of capital. All of STRC’s 2025 distributions came through as ROC, and the issuer has said it expects that treatment to hold for the foreseeable future. ROC means no tax when the payment lands. Instead, each distribution reduces your cost basis in the shares, and only after basis grinds down to zero do further payouts become taxable, at capital-gains rates rather than ordinary income.
Layer a DRIP on top and you get tax-deferred compounding: each payout buys a new lot with its own basis and holding period while shrinking the basis on the old ones. Deferred, not free. You’re trading income tax today for a larger gain at sale, the classification is redetermined every year from the issuer’s actual earnings, and the January 1099 (nondividend distributions, box 3) is the final word. Keep every statement; your broker’s basis tracking is now doing real work.
This isn’t tax advice. ROC treatment is issuer-specific and not guaranteed year to year. Consult a CPA who knows your situation before treating any of this as gospel.
Bitcoin price context
June was the month the framework’s red zone stopped being theoretical. Bitcoin averaged around $63K, opened near $71K, and closed near $58.5K, roughly a 19 percent down month, with record ETF outflows and a hawkish turn from the Fed doing most of the pushing. As I write this it’s back near $64K. In the framework I run, anything below $70K is red alert: defend the existing fleet, no expansion thinking, evaluate the marginal rigs on the rules. That’s exactly where the kill clock discussion above comes from.
The network side was a whipsaw: difficulty fell just over 10 percent at the June 13 adjustment, took back 7 percent on June 27, then cut 5 percent again at today’s adjustment. Hashprice spent late June under $28 per petahash-day and sits near $31.70 as I write this, lifted by that cut. The mood among small operators matches the tape: by one estimate roughly a fifth of miners are unprofitable at these prices, and the conversation has turned to power costs and grinding through. No prediction from me. The zone drives the decisions, and the zone is red.
Tool of the month
mempool.space. It’s where almost every network number in this issue comes from: live difficulty, the recent adjustments, the estimated date and size of the next one, current price. Free, no account, and the same data is available as a clean public API if you ever want to pull it into a spreadsheet. In a month where my whole theme is checking the actual number instead of the narrative, this is the tool that makes that possible in about thirty seconds. When a brief or a thread tells you what the network is doing, this is where you go to see whether it’s true. I keep it pinned. Not sponsored, not affiliated. It’s just where the ground truth lives.
One number
$64,266. My hosting-only cost basis per bitcoin mined in June, against an average spot around $63,000. For the first time, the operation mined above market: the spread that usually carries these issues inverted. But hosting is deductible, and after the roughly 32 percent shield, the effective after-tax basis lands near $43,700 per coin. Run both numbers on your own rig. The first tells you the market’s mood. The second tells you whether the strategy is working.
Difficulty Adjusted is a monthly newsletter for small operators running hosted Bitcoin mining. Strategy, tax mechanics, and the math behind treating mining as a tax-advantaged bitcoin accumulation strategy. Subscribe at difficultyadjusted.io.
Mining Tracker — $79
The spreadsheet I run my own operation from, generalized for any 1–6 rig hosted miner. Download on Gumroad →
This is not financial or tax advice. Consult a CPA for your specific situation.