Back to Documentation

ASIC Fleet Analytics

Historical profitability analysis across all Bitcoin mining hardware

Contents

Overview

ASIC Fleet Analytics provides a comprehensive historical analysis of Bitcoin mining hardware profitability across all ASIC models ever released. Unlike forward-looking unit economics (which project future returns), fleet analytics looks backwards to understand:

  • How long each ASIC model remained profitable after release
  • Which models have never been unprofitable
  • Industry-wide patterns in mining hardware economics
  • The relationship between energy costs and hardware longevity
  • Manufacturer performance comparisons
Key Insight

This analysis uses actual historical network hashrate, BTC price, and difficulty data to compute what each ASIC's daily profit would have been on every day since its release.

Daily Profit Calculation

For each ASIC model and each day since its release date, we calculate:

Daily BTC Mined $$\text{BTC}_{daily} = \frac{H_{asic}}{H_{network}} \times 144 \times R_{block}$$
Where
$H_{asic}$
ASIC hashrate in TH/s
$H_{network}$
Network hashrate on that day in TH/s
144
Average blocks per day
$R_{block}$
Block reward on that day (accounts for halvings)
Daily Revenue (USD) $$\text{Revenue}_{daily} = \text{BTC}_{daily} \times P_{btc}$$
Daily Energy Cost (USD) $$\text{Energy}_{cost} = \frac{W_{asic}}{1000} \times 24 \times \frac{E_{cents}}{100}$$
Where
$W_{asic}$
ASIC power consumption in watts
$E_{cents}$
Energy price in cents per kWh (user-adjustable)
Daily Profit $$\text{Profit}_{daily} = \text{Revenue}_{daily} - \text{Energy}_{cost}$$

Block Reward Schedule

The calculation accounts for Bitcoin's halving schedule:

Period Block Reward Halving Date
Genesis - Nov 2012 50 BTC —
Nov 2012 - Jul 2016 25 BTC November 28, 2012
Jul 2016 - May 2020 12.5 BTC July 9, 2016
May 2020 - Apr 2024 6.25 BTC May 11, 2020
Apr 2024 - Present 3.125 BTC April 19, 2024

Calendar Date Analysis

This view shows how profitable ASICs were on each calendar date across the fleet. For every date, we aggregate:

  • Average profit — Mean daily profit across all ASICs active on that date
  • 2.5th percentile — Lower bound of the 95% confidence interval
  • 97.5th percentile — Upper bound of the 95% confidence interval

This reveals how market conditions (BTC price, network hashrate) affected mining profitability across the entire hardware landscape over time.

Chart Interpretation

The shaded 95% band shows the range of profitability across different ASIC models. A narrow band indicates all ASICs had similar profitability; a wide band indicates significant disparity between efficient and inefficient hardware.

Days Since Launch Analysis

This view normalizes all ASICs to "day 0" = release date, showing the typical profitability lifecycle of mining hardware:

  • Day 0-100: Peak profitability (newest, most efficient)
  • Day 100-500: Declining profitability as network grows
  • Day 500+: Approaching or past break-even

The aggregated view reveals the "half-life" of mining hardware profitability—how quickly the average ASIC becomes unprofitable after release.

Zero-Cross Point

The zero-cross day is a key metric: the first day (since launch) when the average daily profit across all ASICs becomes negative.

Zero-Cross Definition $$d_{zero} = \min\{d : \bar{P}(d) < 0\}$$

Where $\bar{P}(d)$ is the average profit at day $d$ since launch.

Energy Price Dependency

The zero-cross point is highly sensitive to the energy price assumption. At 6¢/kWh, the average ASIC may remain profitable for 800+ days. At 12¢/kWh, this might drop to under 400 days.

Always-Profitable ASICs

These are ASIC models where every single day from release to present has had positive profit at the given energy price. They represent the most resilient mining hardware in the industry.

Characteristics

Always-profitable ASICs are typically: (1) recently released, (2) highly efficient (<25 W/TH), (3) released during favorable market conditions (high BTC price / low difficulty growth).

The calculation is:

Always Profitable Condition $$\forall d \in [0, d_{today}]: P_m(d) > 0$$

Where $P_m(d)$ is the profit for model $m$ on day $d$ since its release.

Manufacturer Analysis

The manufacturer breakdown shows profitability statistics across each brand's product line:

Metric Description
min Shortest profitable period for any model
q1 25th percentile (first quartile)
median 50th percentile (typical model)
q3 75th percentile (third quartile)
max Longest profitable period for any model
mean Average days profitable

This analysis helps identify which manufacturers consistently produce hardware with longer profitable lifespans.

Cost to Build Infrastructure

The "Cost to Build 1 EH/s" calculation estimates infrastructure costs by efficiency cohort:

Cohort Efficiency Range Typical Pricing
Latest Gen ≤20 W/TH ~$50/TH
Current Gen 20-30 W/TH ~$30/TH
Previous Gen 30-50 W/TH ~$15/TH
Legacy >50 W/TH ~$5/TH
Cost per EH/s $$\text{Cost}_{EH} = P_{TH} \times 1{,}000{,}000$$

Where $P_{TH}$ is the median price per TH/s in the cohort.

Data & Caching

Computing profitability for all ASIC models across all historical dates is computationally intensive. The system uses a multi-tier caching strategy:

Disk Cache (24-hour TTL)

The full profitability DataFrame is cached to disk keyed by energy price. Located in data/cache/asic_fleet_*.cache.pkl.

Memory Cache (1-hour TTL)

API responses are cached in memory using the @memoized decorator with a 1-hour TTL for faster subsequent requests.

HTTP Cache (10-minute TTL)

Flask-Caching provides an additional HTTP-level cache with 10-minute TTL for API endpoints.

Cache Management

Users can clear the cache via the "Clear Cache" button on the Fleet Analytics page. This forces a full recalculation on the next data request—useful after updating the underlying ASIC or price data.

Performance Characteristics

Operation First Load Cached
Full profitability compute 30-60 seconds <1 second
Gantt chart render 5-15 seconds <1 second
Summary stats 1-2 seconds <100ms