Estimating the Energy Impact of the Binance Smart Chain

Since we are using blockchain and web3 to address climate change, we need to come clean about the climate impact of the blockchain itself. You’ve probably already heard about the huge energy use of Bitcoin, which has caused most environmentalists to label blockchain as an evil technology. I’ve certainly gotten some hate mail myself for even suggesting that blockchain could be used for climate change.

My answer to this has always been that the energy use of bitcoin is not because it’s a blockchain but because it uses a Proof of Work (POW) consensus algorithm. Because we do not use POW blockchains, our energy use should be negligible. For example, in the past we’ve used xDai (now Gnosis), a Layer 2 Ethereum blockchain that uses a Proof of Stake (POS) consensus algorithm whose reported energy footprint is much smaller than that of even a credit card transaction.

Recently, though, an analysis by Digiconomist pointed out that while Layer 2 Ethereum blockchains, in their case Polygon, do not use much energy themselves, they still rely on the Layer 1 Ethereum mainnet blockchain for essential services. Because the Layer 1 Ethereum mainnet uses POW consensus, the total energy and climate impact of even Layer 2 Ethereum blockahins are much higher. For example, the analysis showed that Polygon’s climate impact was 430 g CO2e per transaction, versus an original estimate of 0.2 g CO2e.

Fortunately there are Layer 1 Proof of Stake (POS) networks available now, and this analysis by the Carbon Crypto Ratings Institute (CCRI) shows how to estimate their energy impact.  It is a fairly straightforward calculation:

  • Determine the hardware used
  • Estimate the power usage of the hardware running the network
  • Multiply by the number of validator nodes on the network
  • Divide by the number of transactions

UPDATE (2022-05-03): This analysis is similar to that of “Energy Footprint of Blockchain Consensus Mechanisms Beyond Proof-of-Work” and Hedera’s analysis of its emissions footprint, but Hedera’s analysis also includes its testnet nodes. This is correct. The energy use of testnet nodes should be included, and the testnet transactions should not be included, leading to higher total and per transaction energy use.

While a precise calculation of energy impact would require actual data about the computing nodes, this methodology gives a reasonable estimate for comparing different networks. According to this analysis, Solana is the most energy efficient Layer 1 POS network.  The problem with using Solana, though, is that we already have our code in Solidity for Ethereum networks.  It would take some work (and energy) to rewrite it.  Once Ethereum 2.0 comes out using POS consensus algorithm, that may well prove unnecessary.

Meanwhile, there is another Ethereum-compatible Layer 1 POS network that we could use now, the Binance Smart Chain. It was not part of the analysis which covered Solana, but we can estimate its energy using the same methodology:

  • Hardware: See Binance Smart Chain Validator FAQs. The 8 core, 16 GB CPU is most similar to option 4 on Table 4 of page 11 of CCRI’s report, which is a 6 core, 32 GB CPU.
  • Power usage of the hardware: The CCRI report calculated that option 4 hardware used between 27 – 70 W per node. We could assume 50 W per node for the Binance Smart Chain.
  • Nodes: There are 21 nodes for validating the Binance Smart Chain (again from Binance Smart Chain Validator FAQs.) For now, let’s assume they all use 50 W per node, or 438 kWh/year (50*24*365).
  • Total energy use is then 21 node * 438 kWh/year, or 9.198 MWh per year. This is much smaller than any of the other Layer 1 POS networks in the CCRI report, which ranged between 70 MWh (Polkadot) and 1,968 MWh (Solana). Meanwhile Ethereum Layer 1 uses 106 Terawatt hours per year, and Bitcoin over 177 Terawatt hours. (Each Terawatt is 1,000,000 Megawatts.)
  • Per transaction: The Binance Smart Chain runs about 6 million transactions per day Its daily energy use is 21 nodes * 50 W * 24 hours = 25,200 Wh, or about 0.004 Wh per transaction. This is even lower than the 0.17 Wh per transaction CCRI estimated for Solana, the lowest Layer 1 POS network in its sample. It is a lot better than Visa credit card processing, which uses 1.78 Wh per transaction, according to Gnosis/Xdai’s Energy Consumption Statistics. \
  • UPDATE (2022-05-03): These energy consumption numbers should include the testnet nodes. Solana says it has 200 testnet nodes in addition to the 1,015 mainnet nodes discussed in the CCRI paper, so its energy use should be about 20% higher, or 2,362 MWh per year and 0.2 Wh per transaction. We don’t know how many nodes there are in the Binance Smart Chain. If we assume it’s also 21 nodes, then the Binance Smart Chain’s energy use and per transaction energy use should be doubled to 50,400 Wh and 0.008 Wh per transaction.
  • To correctly convert this into climate impact, we would need to know where and when the energy was used. The best we could do for now is to use the EIA’s statistics for the average emissions of US electricity, which gives us 1.6 g CO2 per transaction.

If at some point we could know the actual hardware of each computing node and which electricity grids they use, we could come up with a more precise calculation of the energy and climate footprint of our blockchain use. We could even use time- and grid-matched Renewable Energy Certificates to procure renewable energy for it. For now, though, this analysis is a starting point that helps us compare the different blockchain options.

Disclosure/Disclaimer: I do own some SOL (Solana) and xDAI coins.  It did not influence this analysis.  I’m not recommend anybody invest or trade in them.  Cryptocurrencies are very volatile and risky, so do your own research and stay within your risk tolerance. 

One thought on “Estimating the Energy Impact of the Binance Smart Chain

Comments are closed.