Cambridge’s Digital Mining Industry Report put Bitcoin’s annual electricity use at about 138 TWh. That is roughly 0.5 % of global consumption. The live Cambridge Bitcoin Electricity Consumption Index has been running higher on an annualized basis in 2026. The exact number moves with hashrate and machine efficiency, but the order of magnitude does not.

This is not a rounding error. It is continuous demand. The network does not produce goods, services, or cash flow. It requires power to keep existing as a ledger. In a year when Amazon’s trailing free cash flow went to an outflow of $42 billion and Morgan Stanley lifted U.S. data-center power demand another 43 %, that distinction matters.

I have written about the physical constraint before. GPUs arrive faster than transformers and permits. Stranded compute sits dark. Electricity is the payment medium. Bitcoin mining is one of the loads that competes for the same scarce megawatts. Some listed miners have already started converting sites to AI hosting precisely because an energized connection is worth more than the hash it used to run. That is the market admitting the hierarchy.

The price can go up again. The energy cost does not disappear when it does.

The 70/30 portfolio does not include it. We still prefer cash flow and the companies that own or produce the power the new demand actually needs. Crypto can function as a speculative store of value for some people. It does not generate the free cash flow or the grid capacity that the rest of the balance sheet is trying to own.

The headline risk is volatility. The quieter one is that the network keeps consuming a non-trivial amount of the scarce input while producing none of the output the capital is being spent to create.