By February 2026, Bitcoin hashprice had fallen to approximately $34/PH/s - down 35% year-over-year, the weakest level since the China mining ban. BTC dropped 31% from its October 2025 all-tim
By February 2026, Bitcoin hashprice had fallen to approximately $34/PH/s - down 35% year-over-year, the weakest level since the China mining ban. BTC dropped 31% from its October 2025 all-time high, difficulty held, and hashprice fell faster than either. Network hashrate followed, declining roughly 12% from its November peak.
Each time the floor drops, the same conversation starts among operators: which pool am I on, and is it making this worse.
When margins are wide, suboptimal pool decisions are easy to absorb. When hashprice compresses to $35, they're not. Stale shares you tolerated at $90K BTC cut into weekly payouts. Payout delays become a cash flow problem. Downtime logged as "within range" hits margins that no longer have room for it.
Most operators track $BTC price. Fewer model hashprice - revenue per petahash per day - which is what actually tells you whether your fleet is covering costs.
When BTC drops 25% and difficulty stays elevated from the prior epoch, hashprice can fall 30-40% in the same window. That spread is where operations that looked profitable on paper go underwater.
At $0.055/kWh, a fleet of S19 Pro units (100 TH/s, ~34.5 W/TH) on stock firmware is already on thin margins at mid-cycle hashprice. A 35% hashprice compression over two difficulty periods doesn't leave room for a 0.8% stale rate or inconsistent pool uptime. Numbers that feel like rounding errors in a bull market become the actual margin.
Operators with sub-$0.04/kWh power, current-gen S21 or T21 fleets, and tuned firmware absorb the compression. Everyone else starts making harder decisions.
Hashrate doesn't exit the network evenly during drawdowns. It exits at the margin.
The first machines offline are older-gen ASICs - S9s, early S17s, anything on stock firmware - in high-cost jurisdictions or on legacy hosting contracts. When mining revenue drops below the electricity and maintenance threshold, halting is the rational move.
Pools don't lose hashrate uniformly either. Pools with more retail and small-scale miners see disproportionate exits. Foundry USA, AntPool, and F2Pool have enough institutional clients to maintain more stable floors. According to b10c's April 2025 centralization analysis, the five largest pools - Foundry (30%), AntPool (19%), ViaBTC (14.5%), F2Pool (10%), and MARA Pool (5%) - control roughly 78.5% of network hashrate. More critically, just six block template producers mine over 95% of all blocks.
Source: ‘Network share of the current top five mining pools’ by b10c
Stratum V2 is directly relevant here. The protocol shifts block template construction from pool to miner - individual miners decide which transactions to include, not the pool. Beyond reduced data transmission, it's a structural shift in where censorship resistance actually sits. A network where three pools can coordinate to exclude specific transaction types is a different network than one where template construction is distributed. Most large pools haven't moved on this. Braiins has.
When margins compress, miners read pool documentation more carefully.
FPPS vs. PPLNS isn't academic during a downturn. Under FPPS, the pool assumes variance risk and pays a fixed per-share rate inclusive of transaction fees - predictable, easier to model against energy invoices. Under PPLNS, payouts track actual block findings and fee revenue, which outperforms during high mempool activity but introduces variance smaller operations can't absorb.
Liquidity architecture matters as much as payout model.
Exchange-backed pools - Binance Pool, ViaBTC, WhitePool - offer a structurally different workflow. Without exchange integration: rewards hit a wallet, you initiate a withdrawal, wait for confirmations, transfer to an exchange, then hedge or convert. For a 5 PH/s operation managing weekly USD-denominated energy invoices, that's three to four steps and 30–90 minutes of execution risk on a position moving against you.
Pool-to-exchange-native architecture removes those steps. Rewards settle inside a platform where hedging, conversion, or holding are immediate. For operations actively managing BTC exposure, that workflow difference doesn't show up in a fee comparison table.
A firm with a dedicated OTC desk won't need it. A mid-size operation running its own treasury against real-time energy costs probably will.
Bull markets mask pool inefficiencies. A 0.5% stale rate or periodic payout delay doesn't affect decisions when total revenue is high. When hashprice drops 40%, those same numbers are a measurable share of remaining margin.
Orphan block exposure, Stratum latency, server geographic distribution, firmware compatibility with Braiins OS+, VNish, or stock variants - these separate infrastructure built for sustained load from infrastructure that degrades when tested.
Luxor's hashrate index is the most-cited infrastructure benchmark in North American mining, and their transparency reporting has set the standard for pool-side disclosure. Braiins' FPPS+ model with open firmware integration is validated by operators running custom ASIC configurations - not marketing, a stack with a real track record.
Exchange-backed entrants including WhitePool are being stress-tested on this now. Uptime, stale rates, and payout reliability under volatility - not favorable conditions - is the live question. No pool has a clean record on every metric. What matters is how issues are handled, how they're communicated, and whether the infrastructure justifies the fee.
Post-halving, with block subsidy at 3.125 BTC and fee revenue still inconsistent, pool economics don't forgive loose evaluation.
Payout model. FPPS for treasury predictability; PPLNS if you're large enough to absorb variance and want upside in high-fee environments.
Foundry USA and AntPool dominate by hashrate. F2Pool has the longest track record. Luxor and Braiins serve operators who prioritize transparency and firmware flexibility. Exchange-integrated pools like WhitePool target operators who want mining economics and treasury management in the same place.
The right pool depends on power cost, fleet composition, treasury structure, and scale. What doesn't change: operators who evaluate pools on verified specs and operational history hold margins when everyone else is calculating whether to keep the lights on.
Disclaimer: This is not financial or investment advice. Do your own research before making any decisions. Use at your own risk.
