The Digital Economy Is Running Into the Physical World

Bitcoin has always had a useful habit: it makes the supposedly virtual economy look stubbornly physical.

JPMorgan recently estimated the cryptocurrency’s average production cost at roughly $85,000. According to The Block, citing the bank’s analysts, Bitcoin has spent around 280 days below that level. A sustained move above it could ease pressure on miners forced to sell coins to cover their costs.

For crypto investors, that is interesting enough. But the more important story lies elsewhere.

Bitcoin miners are now competing with a new customer for the same electricity, land and data-center infrastructure: artificial intelligence. In some locations, switching capacity from cryptocurrency mining to AI can produce higher and more predictable returns per megawatt.

The significance of this goes well beyond Bitcoin.

For years, digitalization has been described as though economic activity were gradually escaping the physical world. Money becomes electronic. Assets are tokenized. Commerce migrates to platforms. Artificial intelligence substitutes for human labor. Data vanishes into the cloud.

The vocabulary itself suggests dematerialization.

The reality is almost the reverse.

The more digital the economy becomes, the more dependent it is on physical infrastructure.

Not abstractions, but power stations. Gas. Uranium. Copper. Aluminum. Transformers. Transmission lines. Cooling systems. Water. Semiconductor plants. Data centers. Fiber-optic cables. Satellites. Ports. And the people, companies and governments capable of building, maintaining and defending them.

Bitcoin merely exposed this dependence earlier than most digital industries.

Mining cannot disguise its appetite for electricity. When power becomes more expensive, mining economics deteriorate. When more profitable users appear, miners lose access to capacity. The supposedly virtual asset quickly runs into the hard constraints of energy, equipment and land.

That does not make JPMorgan’s $85,000 estimate some kind of fundamental or “true” value for Bitcoin. Mining costs vary with electricity prices, hardware efficiency, network difficulty and a host of other assumptions.

But the figure is useful for a different reason.

It points towards a problem that is spreading across the digital economy.

The International Energy Agency expects global electricity consumption by data centers to more than double by 2030, reaching about 945 terawatt-hours. That is more electricity than Japan currently consumes in a year.

Artificial intelligence is expected to be the main driver.

In America, the IEA projects that data centers could account for almost half of the increase in electricity demand between now and the end of the decade.

This creates a rather different kind of digital competition.

Steel mills, chemical plants, railways, households, electric vehicles, Bitcoin miners, cloud providers, AI companies and governments increasingly want the same thing.

A megawatt.

The digital economy was once regarded as something that sat on top of the energy system: clever, weightless and scalable. It is now becoming one of the largest new sources of demand within that system.

And producing electricity is only half the problem.

It must also reach the customer.

In its Electricity 2026 report, the IEA identifies power grids as one of the most important emerging bottlenecks in the global energy system. More than 2,500 gigawatts of new generation, storage projects and large electricity consumers are waiting for grid connections.

The world currently invests around $400 billion a year in electricity networks. The agency reckons that figure will need to rise by roughly 50% by 2030.

The timing mismatch is awkward.

A data center may take one to three years to build.

A major grid expansion can take five to fifteen.

That is the first great constraint of the new digital economy.

Software moves faster than steel.

An AI service can be deployed in weeks. Software platforms can scale in months. A token can be created in seconds.

A nuclear reactor cannot.

Nor can a turbine, transformer, transmission line or submarine cable.

Digital systems can expand at software speed. Their physical foundations cannot.

Energy is therefore becoming a natural ceiling on digital expansion.

There is a useful, if imperfect, historical analogy here.

For centuries governments struggled with the relationship between financial claims and the physical resources supporting them. Roman emperors could debase coins by reducing their silver content. Medieval rulers did much the same.

Modern monetary systems work very differently. Central banks do not operate like imperial mints. Bank deposits, stablecoins and tokenized bonds are not denarii with less silver in them.

Still, the underlying question has not disappeared.

How many financial claims can an economy support relative to the physical system required to service them?

Money can be created.