Ethereum co-founder Vitalik Buterin compared the cost structures of Bitcoin mining and AI inference in an Oct. 17 X post. Electricity makes up 45.9% of mining costs in the table he shared, while computing capital expenditure makes up approximately 76% of the inference estimate.
Vitalik Buterin described Bitcoin mining costs as split about 50/50 between computing capital expenditure and electricity.
Electricity takes 45.9% of the mining estimate
The mining table puts electricity supplied directly to computers at $40,600 per BTC, or 43.3% of the total. Other electricity expenses add $2,400, or 2.6%.
Together, the two electricity categories come to $43,000 per BTC and 45.9% of the listed cost. Computing capital expenditure contributes approximately $37,600 per BTC, around 40%, and covers hardware, cooling, and power equipment, including financing.
The remaining categories are labor at $3,200 per BTC, facility rent at approximately $2,500, and other expenses at roughly $7,400. The listed amounts sum to an estimated $93,700 per BTC, but that figure describes the shared table and does not set a uniform production cost for every miner.
AI inference leans on computing capital expenditure
The AI table uses dollars per GPU-year. It places computing capital expenditure at approximately $19,900 per GPU-year, or 76% of the total, and defines the category the same way as the mining table.
Electricity supplied directly to computers accounts for $1,050 per GPU-year, or 4%. Other electricity adds $210, or 0.8%, which brings the combined electricity share to 4.8%.
Buterin wrote that AI inference is "almost completely dominated by compute capex". Because one table measures costs per Bitcoin and the other per GPU-year, their dollar totals do not offer a direct comparison of the price of performing the same task.
Source: crypto.news
Trading involves risk.
Risk Disclaimer
