Starlink and the Next Rural Electrification

There is something slightly strange about watching a Starlink dish connect to the internet.

You plug it in, point it at the sky, and a few minutes later a house in a place where running fibre would make little economic sense is communicating with the rest of the planet at broadband speeds.

The interesting part is not really the satellite. It is what happens when a network reaches places that terrestrial infrastructure would have reached much later, or perhaps never.

Starlink describes its system as a constellation of thousands of low Earth orbit satellites. Its published service specifications put typical land latency at 25–60 ms and download speeds between 25 and 220 Mbps, with actual performance varying by plan, location, and congestion. Those are service targets and ranges, not a guarantee for every terminal.

Those numbers are impressive engineering. The more interesting number is the number of people for whom the alternative was waiting.

The last few kilometres are the expensive part

Infrastructure has an awkward property: the places that need it most are often the places where it is hardest to justify building.

A city can support fibre because thousands of customers live along the same streets. A dense neighbourhood can support a cellular tower because the cost of the tower and its backhaul is distributed across many subscribers.

A village in a remote region is different. The economics become dominated by the last few kilometres, then the last few kilometres after that.

This is one reason the history of electricity is useful here.

In 1930, fewer than 10% of American farm households had electricity. Urban America was already living in an increasingly electrified world while much of rural America was not. The problem was not that electricity had not been invented. The problem was that extending the grid to sparsely populated areas was difficult to justify economically.

The Rural Electrification Administration (REA) attacked that constraint through financing and local organisation. The 1936 Rural Electrification Act enabled loans to rural organisations and cooperatives so they could build distribution networks from the existing grid. In its first five years, the REA provided more than $227 million in loans and more than doubled the number of electrified farms. The historical rollout and its effects are documented in this NBER paper.

The important mechanism was not simply “build more power plants”. It was to change who could finance and coordinate the distribution network.

The consequences were larger than having electric lights. Electricity changed agriculture, household work, the machines that could be used, and the businesses that could operate in rural areas. Research on the historical rollout found that counties receiving electricity earlier experienced economic growth that persisted for decades after the country itself had become fully electrified. The measured long-run effects include population, employment, income, and property values.

The infrastructure did not merely improve the existing economy. It changed the set of things that were possible.

Starlink changes the network topology

Starlink has a similar shape, although the institutional model is very different. The REA was a public financing programme; Starlink is a privately operated communications network. The similarity is in the infrastructure problem they address:

How do you give modern infrastructure to people for whom conventional infrastructure is economically inconvenient?

Terrestrial broadband usually grows outward from population centres. It needs a route, local access equipment, backhaul, maintenance, and enough nearby demand to pay for them. A remote customer is expensive because the network must be extended to reach that customer.

Low Earth orbit reverses part of that arrangement. The expensive shared segment is placed in orbit, while the local segment is a user terminal, a power supply, and a view of the sky. The satellite forwards traffic to a gateway or to other satellites before it reaches the terrestrial internet. The details vary as the constellation and ground network evolve, but the economic effect is the important part: the connection no longer requires a dedicated fibre route all the way to the house.

That makes a remote house, farm, research station, village, ship, aircraft, or isolated industrial site a network endpoint without first requiring a profitable terrestrial market in that location.

This is why I find the technology more interesting as infrastructure than as a gadget. The satellite is the implementation. The consequence is that geography becomes less important in the cost of reaching a new subscriber.

Broadband creates an option space

Electricity was one of the defining general-purpose technologies of the twentieth century. The internet is one of the defining general-purpose technologies of the twenty-first.

Someone who has electricity can use technologies that did not exist when the power line was built. The same is true of internet connectivity. A village connected today does not merely receive websites. It receives access to online education, software, remote work, financial services, scientific literature, telemedicine, global markets, programming tools, and communication.

The last category is the important one: whatever has not been invented yet.

We are bad at estimating infrastructure by looking only at the applications that exist when it is deployed. A power line built in 1936 could not have been evaluated by asking whether it would eventually power a microwave oven, a computer, or a modern factory. Likewise, the value of giving a remote community broadband cannot be estimated by listing today's websites.

The better model is an option space. Connectivity makes more actions feasible, and some of those actions will have value that is invisible in a subscriber count.

A teenager in a remote village with broadband can discover a university lecture, an open-source project, a programming language, a scientific paper, a business opportunity, or a community with the same obscure interest. Most of those events will never appear in a Starlink metric. That does not make them insignificant; it makes them difficult to forecast.

The analogy has boundaries

The rural-electrification analogy is useful only if its limits stay visible.

Starlink does not remove the need for terrestrial networks. It depends on ground stations, fibre backhaul, spectrum, power, terminals, maintenance, regulation, and a clear enough view of the sky. Capacity is shared: a satellite can cover a large area, but its available throughput is finite and is divided among active users. Rain and obstructions can degrade the link, and the terminal and monthly service still cost money.

The system also cannot make geography irrelevant in every sense. It changes the cost curve for the access link. It does not make a remote location cheap to staff, power, or supply, and it does not guarantee that a connection will create useful local institutions or affordable devices.

Those constraints matter because infrastructure is a system, not a single component. A dish that can see a satellite is not the same thing as a community with reliable, useful access. The technology creates an opportunity; education, policy, local networks, and economic activity determine how much of that opportunity is used.

The useful comparison

The digital divide may gradually shrink as countries become wealthier and terrestrial infrastructure expands. But some places may also skip part of that sequence. They do not need to wait for fibre to reach them before participating in the broadband economy, and they do not necessarily need to reproduce the infrastructure path taken by cities.

That is what makes Starlink feel, to me, somewhat like rural electrification. Not because a constellation of satellites is literally equivalent to the Rural Electrification Administration. It is not. The similarity is more fundamental: both change the economics of extending a general-purpose network to low-density places.

The answer in the 1930s was to change the financing and organisation of grid distribution. The answer in the 2020s is, in part, to change the topology of the communications network.

If Starlink succeeds at scale, its most important achievement may not be that people in remote places get faster internet. It may be that millions of people who were previously outside the information economy gain a practical route into it.

The most interesting infrastructure is often the infrastructure whose second-order effects are impossible to predict.

A power line looks like a wire until you ask what people can do after you give them electricity.

A satellite dish looks like an antenna until you ask what people can do after you give them the internet.

I expect all of the benefits described above to be greatly amplified as the technology improves over the coming years. Today's first orbital launch of Starship, on September 28, 2026, carried the first batch of 26 Starlink V3 satellites into orbit. The new satellites are a step improvement in capability, and contemporary launch coverage describes the historic mission and its objectives in more detail.

The following video goes deeper into the technical details of this technological marvel.

I personally consider the Starlink constellation one of the great modern wonders of human civilisation, up there with electricity, the transistor, the integrated circuit, and ultraviolet lithography.