Mining Hardware & Pools

ASIC hardware specs and manufacturers, GPU mining, mining pool reward schemes (FPPS, PPS+, PPLNS), pool centralisation risk, solo mining variance, hosting vs self-mining, and ASIC market signals.

Course 59: Mining Hardware & Pools

Blockchain & Mining Track · 28 min read · Intermediate

Bitcoin mining hardware has undergone more than a million-fold efficiency improvement in fifteen years. The first Bitcoin blocks were mined on general-purpose CPUs running on personal computers in 2009, earning 50 BTC per block at effectively zero marginal electricity cost. By 2011, GPU mining had replaced CPU mining entirely. By 2013, ASICs — application-specific integrated circuits designed to do nothing except compute SHA-256 hashes — had rendered GPU mining on Bitcoin permanently uneconomic. Today, a single modern ASIC consumes approximately 3,500 watts and performs 335 trillion hashes per second, and the entire network collectively performs approximately 600 exahashes per second (600,000,000,000,000,000,000 hashes/second). The hardware that won that computation race, the pool infrastructure that distributes mining rewards across thousands of participants, and the economic logic connecting them to Bitcoin's price and your trading decisions are the subject of this course. It extends the mining economics of Course 58 to the operational layer.

ASIC Hardware: How It Works

An ASIC (Application-Specific Integrated Circuit) is a chip designed to perform a single computational function at maximum efficiency. Bitcoin mining ASICs compute SHA-256 double hashes — the specific hash function in Bitcoin's proof-of-work algorithm — and nothing else. They cannot be reprogrammed to mine Ethereum (which uses Keccak-256), run general computing tasks, or play games. This specialisation is what makes them orders of magnitude more efficient than general-purpose processors for their specific task.

Key ASIC hardware metrics:

  • Hashrate (TH/s): terahashes per second. The number of SHA-256 double hashes the ASIC can compute per second. Current top-tier hardware: 200–335 TH/s per unit.
  • Power consumption (watts): the electrical power draw. Ranges from ~2,800W for older efficient models to ~6,800W for high-performance air-cooled units.
  • Efficiency (J/TH): joules per terahash — the most critical metric. Lower is better. This measures how much electricity you consume per unit of mining work. Current generation efficiency: 15–20 J/TH for leading models. Previous generation (circa 2021): 30–40 J/TH. Earlier generation (circa 2019): 50–80 J/TH.

Why efficiency matters: at $0.07/kWh electricity cost, the difference between a 20 J/TH machine and a 40 J/TH machine operating at 100 TH/s is $0.048/hour — $420/year per machine. At industrial scale with thousands of units, efficiency differences directly determine competitive viability. During bear markets, only the most efficient hardware remains profitable; inefficient machines are shut down and sold for scrap value.

Major ASIC Manufacturers

Three companies dominate Bitcoin ASIC production, each with a product line covering the current and previous mining generations:

  • Bitmain (Antminer series): the dominant manufacturer by market share since 2013. Produces the Antminer S-series (flagship Bitcoin miners) and the L-series (Litecoin/Scrypt miners). The S21 Pro (2024) operates at 234 TH/s at ~15 J/TH. Bitmain is a private Chinese company with significant market power — its manufacturing capacity, new product release timing, and hardware availability directly affect the trajectory of Bitcoin hashrate globally.
  • MicroBT (Whatsminer series): Bitmain's primary competitor. Products include the M50S++ and M66S+ series. The M66S+ (2024) achieves 298 TH/s at approximately 18 J/TH. MicroBT has gained market share from Bitmain since 2020, providing competitive diversification in the hardware supply chain.
  • Canaan (Avalon series): the original Bitcoin ASIC manufacturer (Canaan shipped the first commercial ASIC in 2013), now a publicly traded company on Nasdaq (CAN). Avalon hardware (A15 series as of 2024) is competitive but generally perceived as second-tier to Bitmain and MicroBT in efficiency per generation.

GPU Mining: Where It Still Exists

General-purpose GPU mining was rendered uneconomic on Bitcoin by ASICs in 2013. However, GPU mining persists for algorithms that are ASIC-resistant:

  • Monero (XMR/RandomX): uses the RandomX algorithm, specifically designed to be efficient on CPUs/GPUs and inefficient on ASICs. RandomX periodically changes to maintain ASIC resistance. Monero is effectively the last major proof-of-work coin that can be mined profitably on commodity hardware (CPU/GPU).
  • Etchash (Ethereum Classic): Ethereum migrated to PoS in September 2022, eliminating GPU mining on ETH overnight. Many GPU miners shifted to Ethereum Classic (ETC) and Ergo. The combined market cap and hashrate of these alternatives is far smaller than Ethereum was, making GPU mining significantly less profitable post-merge.
  • Kaspa (KHeavyHash): an emerging PoW cryptocurrency that has attracted GPU and ASIC mining since its rapid hashrate growth in 2023.

From a trader's perspective, GPU mining demand is relevant as a marginal demand signal for NVIDIA and AMD cards (affecting their valuations) and as an indicator of speculative mining activity in smaller PoW coins. The collapse in GPU mining revenue post-Ethereum Merge was partly responsible for the GPU oversupply and price drops in the consumer graphics card market in 2023.

Why Mining Pools Exist: Variance Reduction

Bitcoin mining is a probabilistic process. The probability that a single mining unit finds the next block is approximately proportional to its share of total network hashrate. A miner controlling 0.0001% of total Bitcoin hashrate — say, two Antminer S21s — would on average expect to find one block every 27 years. They might find one tomorrow or wait 100 years — purely random. This variance is economically unworkable: a small operation cannot sustain 27 years of electricity costs on the hope of an occasional block reward.

Mining pools solve this by aggregating thousands of miners' hashrates. Each contributor submits "shares" (work proving they are contributing their proportional effort). When any pool member finds a valid block, the reward is distributed among all contributors proportional to their submitted shares. The expected time between payouts for a small miner drops from decades to hours or days, eliminating catastrophic variance at the cost of the pool operator's fee (typically 1–2%).

Mining Pool Reward Schemes: Variance ComparisonFPPSFull Pay Per ShareMiner earns fixed rateper valid shareregardless of block foundMiner variance: LOWESTPool takes block riskHigher pool fee ~2-3%PPS+Pay Per Share (+ fees)Fixed rate for subsidyproportional share of feeswhen blocks foundMiner variance: LOWMost common scheme~1.5-2% feePPLNSPay Per Last N SharesRewards proportional toshares in the last N windowwhen a block is foundMiner variance: HIGHERPool variance: LOW~0.9-1.5% fee
Fig 1 — Mining pool reward scheme comparison. FPPS minimises miner variance; the pool bears the risk of unlucky block finding and charges higher fees. PPLNS passes more variance to miners but charges lower fees and discourages pool-hopping (a miner who joins just before a block is found earns less than a consistent contributor). PPS+ is the most common in practice.

Pool Reward Schemes

The primary pool payment schemes determine how risk is distributed between the pool and its miners:

  • PPS (Pay Per Share): the pool pays a fixed amount for every valid share submitted, regardless of whether the pool finds a block. The pool assumes all variance risk. The per-share payment reflects the statistical expected value of a share. Foundry USA, the largest Bitcoin pool, uses a variant called FPPS which includes transaction fees in the fixed payment.
  • PPLNS (Pay Per Last N Shares): rewards are calculated based on a miner's proportional contribution of shares to the last N shares submitted before a block was found. If the pool has a "lucky" streak and finds blocks faster than expected, miners earn more; if unlucky, less. PPLNS incentivises loyalty: miners who leave the pool just before a block is found get lower pay; consistent miners benefit from lucky streaks. Popular on pools like F2Pool and ViaBTC.
  • SOLO mining: the miner keeps 100% of block rewards when they find a block, minus a small pool hosting fee (0.5–1%) for the block coordination infrastructure. Revenue is extremely high variance: a miner with 0.01% of network hashrate would wait ~2.7 years on average between blocks. Generally only rational for very large operators who can absorb multi-month variance on their treasury. Some smaller miners use solo pools as a "lottery" strategy with a small portion of their hashrate.

Pool Centralization Risk

Bitcoin's security rests on no single entity controlling more than 50% of hashrate. Pool concentration is the primary centralisation risk in modern Bitcoin:

  • In July 2014, GHash.IO briefly exceeded 51% of Bitcoin's total hashrate for several hours. No attack was launched, but the episode demonstrated the vulnerability and triggered significant debate about pool centralisation. GHash.IO subsequently voluntarily reduced its hashrate to below 40%.
  • As of 2024, Foundry USA controls approximately 30–35% of Bitcoin hashrate, followed by AntPool (~15-18%), F2Pool (~10-12%), and Binance Pool (~4-6%). No single pool has reached 51% since 2014, but the top three pools collectively control approximately 55–60% of hashrate.
  • It is important to distinguish pools from actual mining entities: pools aggregate hashrate from many independent miners who could switch pools if their pool behaved maliciously. Pool operators themselves typically hold no hardware. The practical 51% attack threat is from a coordinated attack by pool operators — which would destroy the value of the Bitcoin they mined, a strong disincentive. Nonetheless, monitoring pool concentration remains a network health metric tracked by researchers.

Hosting vs Self-Mining

For individual or institutional miners, the choice between self-managed facilities and hosted mining determines both cost structure and operational complexity:

  • Self-mining: the operator owns or leases the physical facility, is responsible for power procurement, cooling, security, and maintenance. Offers maximum cost control but requires significant capital for facility buildout and expertise to manage. Industrial self-miners (Marathon, Riot, CleanSpark) operate proprietary data centres in low-cost power markets (Texas, Wyoming, Alberta, Quebec) with power purchase agreements (PPAs) at $0.03–$0.05/kWh.
  • Hosted mining: a hosting provider manages facility, power, cooling, and maintenance for a monthly hosting fee (typically $0.06–$0.12/kWh all-in) or a fixed fee per machine. The client owns the hardware and receives mining proceeds. Reduces barriers to entry but higher effective electricity cost and counterparty risk on the hosting provider's solvency and reliability.
  • Hash power leasing / cloud mining: purchasing hashrate contracts from a provider without owning hardware. The purchaser receives mining proceeds but has no asset ownership at contract expiry. Historically, many cloud mining operations have been fraudulent or uneconomic. Legitimate cloud mining contracts (NiceHash, Compass) exist but require careful economic modelling to determine whether the implied electricity rate makes them competitive with direct hardware ownership.

Profitability Analysis

Mining profitability depends on five inputs: hardware hashrate (TH/s), hardware efficiency (J/TH), electricity cost ($/kWh), Bitcoin price ($/BTC), and network difficulty (which sets how much BTC each petahash earns per day). Profitability calculators (f2pool.com, asicminervalue.com, nicehash.com/profitability-calculator) solve this system for a given hardware configuration.

The key principle: hardware purchased at peak bull market prices often becomes unprofitable within 6–18 months due to the combination of price decline, difficulty increase from new hardware deployments, and halving-driven reward reduction. ASICs purchased in Q4 2021 (peak bull market) at $10,000+ per unit were operating at large losses by Q4 2022 and were being sold for scrap at $300–$600/unit. Conversely, ASICs purchased during bear market capitulation at distressed prices often have 3–5 year profitable operating lives in the subsequent bull cycle. Hardware timing is a significant component of mining ROI, analogous to the entry timing analysis covered in Course 6. Use the crypto profit and loss calculator to model break-even scenarios and the free crypto tools for hashprice sensitivity analysis before any mining capital deployment.

Implications for Traders

Mining hardware and pool dynamics provide several indirect but useful signals for traders who do not mine themselves:

  • ASIC secondary market prices: distressed ASIC prices (below $1/TH) historically correlate with market bottoms. Recovery in ASIC prices indicates mining industry health and often leads price recoveries by 1–3 months.
  • New hardware generation cycles: when Bitmain or MicroBT announces a new hardware generation with significantly improved efficiency, expect network hashrate to increase substantially 4–8 months later when the units ship. Rising hashrate from more efficient hardware increases mining cost for all participants and often creates difficulty adjustment upward pressure that compresses miner margins.
  • Pool concentration monitoring: if a single pool approaches 40% or above, it is worth noting as a network health risk. Above 50% would be an immediate security concern affecting Bitcoin's fundamental value proposition. Monitor via mempool.space's hashrate distribution page.
  • Mining expansion as a bullish signal: when public miners are expanding their machine count and facility buildout rather than conserving capital, it signals institutional confidence in forward Bitcoin price. Miner guidance in quarterly earnings calls provides a real-time window into how the most informed participants in the Bitcoin supply chain are positioning. Use this alongside the on-chain signals from Course 60.

Summary

Bitcoin mining hardware has evolved from CPUs to highly specialised ASICs with 15–20 J/TH efficiency, produced by three primary manufacturers (Bitmain, MicroBT, Canaan). Mining pools emerged to solve the variance problem inherent in probabilistic block finding, enabling small-scale miners to receive regular, predictable payouts in exchange for a 1–3% fee. Pool reward schemes (FPPS, PPS+, PPLNS) differ in how block-finding variance is distributed between the pool and its miners. Pool concentration — with Foundry controlling ~30% of Bitcoin hashrate — is a monitored network health metric. Hosting vs self-mining trade-offs determine effective electricity cost and operational risk profile. ASIC market prices, new hardware announcements, and miner capacity expansions provide indirect but valuable signals for Bitcoin price trajectory. The complete on-chain analytics toolkit for synthesising mining signals with market structure is assembled in the final course of this track: Course 60: Blockchain Analytics for Traders.

Your Free Crypto Tools for This Course

The concepts introduced in this course connect directly to several of DennTech’s free crypto tools. There is no registration required to use any of them:

← Bitcoin Mining Economics All Courses Blockchain Analytics for Traders →