ASIC Price Index
Fair market value estimation for Bitcoin mining hardware
Overview
The ASIC Price Index estimates the fair market value of any Bitcoin mining ASIC based on its efficiency rating (J/TH). Unlike spot prices for specific models, this index provides a systematic way to value any miner — including models that may not have active market listings.
The tool answers fundamental questions for miners and investors:
- What is my ASIC worth today based on its efficiency?
- How has the price per TH evolved over time for my efficiency class?
- What was an ASIC at this efficiency worth when it launched vs. today?
- How might prices evolve as newer, more efficient models enter the market?
More efficient ASICs (lower J/TH) command higher prices per TH because they generate more profit per unit of electricity consumed. This relationship follows an exponential curve, not a linear one.
Data Source: Luxor Hashrate Index
The ASIC Price Index is built on daily market data from Hashrate Index by Luxor, a leading source of Bitcoin mining market intelligence. Luxor tracks actual ASIC transaction prices from broker networks, OTC desks, and marketplace listings.
Unlike theoretical valuations or manufacturer MSRPs, these prices represent real market transactions — what buyers are actually paying for hardware. The data is updated daily and covers the full spectrum of hardware generations.
Luxor's data has been tracked since January 2018, providing over 6 years of historical pricing across multiple Bitcoin cycles and hardware generations.
Efficiency Buckets
Luxor groups ASIC prices into 5 efficiency buckets based on J/TH (joules per terahash). Each bucket represents a "generation" of mining hardware:
| Bucket | Efficiency Range | Representative Value | Generation | Example Models |
|---|---|---|---|---|
under_19 |
≤ 19 J/TH | ~10 J/TH | Latest Gen | S21 XP, M60S++, S21 Pro |
bucket_19_25 |
19–25 J/TH | ~22 J/TH | Current Gen | S21, M60S, S19 XP |
bucket_25_38 |
25–38 J/TH | ~31.5 J/TH | Previous Gen | S19j Pro, M30S++ |
bucket_38_68 |
38–68 J/TH | ~53 J/TH | Older Gen | S17, M20S, T17 |
above_68 |
≥ 68 J/TH | ~75 J/TH | Legacy Gen | S9, T9, Avalon 741 |
The "Representative Value" is the midpoint efficiency we use for regression modeling.
For example, the under_19 bucket uses 10 J/TH as its representative value
because the most efficient ASICs in this bucket cluster around that efficiency.
The Pricing Model
The relationship between ASIC efficiency and price follows a log-linear (exponential) pattern. More efficient ASICs are exponentially more valuable because small improvements in efficiency yield disproportionately large improvements in profitability.
Mathematical Foundation
For each day in our dataset, we fit a linear regression on the log-transformed prices:
- $\text{Price}_{TH}$
- Price per terahash in USD
- $a$
- Intercept — base log-price at 0 J/TH (theoretical)
- $b$
- Slope — typically negative (higher efficiency = lower J/TH = higher price)
- $\text{Efficiency}$
- ASIC efficiency in J/TH (joules per terahash)
To get the actual price from the model, we exponentiate:
Why Log-Linear?
A log-linear model is appropriate because:
- Prices are always positive — the exponential function ensures no negative predictions
- Percentage changes matter more than absolute changes — going from 20 to 15 J/TH is more valuable than 80 to 75 J/TH
- Market behavior is multiplicative — each J/TH improvement has a proportional (not additive) effect on value
Price Interpolation
The regression model allows us to estimate prices for any efficiency value, not just the bucket midpoints. This is interpolation within the data range and extrapolation outside it.
Suppose on a given day the model parameters are: $a = 3.5$, $b = -0.07$
For an ASIC at 15 J/TH:
$\text{Price}_{TH} = e^{3.5 + (-0.07) \times 15} = e^{3.5 - 1.05} = e^{2.45} \approx \$11.59/\text{TH}$
For an ASIC at 30 J/TH:
$\text{Price}_{TH} = e^{3.5 + (-0.07) \times 30} = e^{3.5 - 2.1} = e^{1.4} \approx \$4.06/\text{TH}$
Prices extrapolated far outside the bucket ranges (e.g., 5 J/TH or 100 J/TH) should be treated with caution. The model is most accurate for efficiencies between 10 and 75 J/TH.
Confidence Metrics
Each price estimate includes confidence metrics to help you assess reliability:
R² Score (Coefficient of Determination)
The R² score measures how well the linear regression fits the bucket data points. It ranges from 0 to 1:
| R² Range | Interpretation | Confidence Level |
|---|---|---|
| > 0.90 | Excellent fit — data points closely follow the trend | High |
| 0.70 – 0.90 | Good fit — reasonable trend with some deviation | Medium |
| < 0.70 | Poor fit — high variance, estimate less reliable | Low |
Data Points (Bucket Count)
The number of efficiency buckets with valid price data on a given day. More data points generally mean a more reliable regression:
- 5 buckets — Full data, highest confidence
- 4 buckets — Very good, one generation may lack liquidity
- 3 buckets — Acceptable, but extrapolation is less reliable
- 2 buckets — Minimum for regression, treat with caution
- <2 buckets — Cannot fit model, no estimate available
As of late 2024, older generation ASICs (38-68 J/TH and above 68 J/TH buckets) often have no market data as they've become unprofitable and stopped trading. This is normal market evolution — only 3 buckets may have active prices.
Time Normalization
A critical insight of the ASIC Price Index is time normalization. The same efficiency rating means very different things at different points in time:
In 2019: 40 J/TH was cutting-edge efficiency. The Antminer S17 Pro at ~40 J/TH was a premium machine selling for $50+/TH.
In 2024: 40 J/TH is old generation. The same efficiency class now sells for $2-3/TH as more efficient ASICs have entered the market.
The Price Index captures this evolution by fitting a new model for each day. This means:
- Historical prices reflect what an efficiency class was worth at that time
- You can track how a specific efficiency depreciated as technology advanced
- Launch price vs. current price comparisons are meaningful
Price Forecasting
The Price Index includes forward-looking price forecasts based on the historical trend of "frontier efficiency" — the best available efficiency at any point in time.
Efficiency Trend Model
ASIC efficiency has improved roughly exponentially over time, following a log-linear trend:
By extrapolating this trend, we can estimate what "latest generation" efficiency will be in the future, and thus what a given efficiency class might be worth as it becomes relatively less efficient.
Price forecasts are inherently uncertain. They assume: (1) efficiency continues to improve at historical rates, (2) the price-efficiency relationship remains stable, and (3) no major market disruptions. Use forecasts as one input among many for decisions.
Limitations
While the ASIC Price Index is a powerful tool, it has important limitations:
What It Captures
- Average market prices for efficiency classes
- Overall price-efficiency relationships
- Historical trends and depreciation patterns
What It Doesn't Capture
- Condition premium/discount — New vs. used hardware at same efficiency
- Brand premiums — Bitmain may command higher prices than other brands
- Bulk discounts — Large orders may have different pricing
- Regional variations — Prices vary by location and import costs
- Warranty/support value — Newer machines have better support
- Liquidity premium — Popular models may command slight premiums
Use the ASIC Price Index as a starting point for valuation, then adjust based on specific factors like condition, brand, quantity, and current market liquidity. The index is most accurate for understanding relative value between efficiency classes and tracking trends over time.
Try the ASIC Price Index
Ready to estimate ASIC values? Use our interactive tools: