The $457 Billion Mirage: Why the Satellite Boom Won’t Connect the Poor — and What Will

Everyone loves a space story. Goldman Sachs has a very interesting new report, that projects a seven-fold expansion in the global satellite market by 2035, worth up to $457 billion. Some 70 000 low-earth-orbit (LEO) satellites could soon wrap the planet in broadband, bringing the internet to the 2.5 billion people who still lack it.

It sounds inspiring — until you run the numbers.

1. The arithmetic that no one’s doing

Even in Goldman’s most optimistic scenario, launching enough satellites to cover half the world would cost $9–18 billion in launch costs alone, assuming 200 kg per satellite at $100–200/kg. Add manufacturing, ground stations, replacements every five to seven years, and debris management, and you’re looking at tens of billions in capital that literally burns up on re-entry.

Compare that to terrestrial options. India’s BharatNet is connecting 250 000 villages with fiber for ~$8 billion — infrastructure that lasts decades. In Africa, tower-sharing and solar-powered base stations cut the cost of mobile expansion to $150 000–300 000 per tower. You can build a lot of ground networks for the price of one orbital dream.

2. A rich person’s technology for poor people’s problems

LEO constellations make economic sense only if high-margin users subsidize low-income ones — ships, aircraft, and oil rigs paying for subsistence farmers. That’s the inverse of how inclusion actually scales.

When Kenya’s M-Pesa and Bangladesh’s Grameenphone took off, it was because the unit economics worked from the bottom up.

Today, a LEO terminal costs $500–600; even at $100, it’s a month’s income for a household earning $3 a day. Monthly service charges run $50–150. No amount of marketing will make that affordable in rural Niger or Odisha.

3. Physics and economics won’t bend for inclusion

Every bit sent to space and back travels a thousand kilometers through vacuum. It will always cost more energy and capital than sending it through fiber or 5G.

Two decades of Viasat and HughesNet in rural America prove the point: better latency doesn’t fix unaffordability. The laws of physics make it impossible for bandwidth from orbit to be cheaper than bandwidth on land.

4. The orbital land-grab nobody’s pricing

The International Telecommunication Union allocates orbits on a “first-filed, first-served” basis.

The tragedy is that this geopolitical chess match creates externalities in orbit — debris, collisions, and exclusion of legitimate development uses such as weather and climate monitoring. Removing a single large piece of debris can cost $200 million. None of that is in anyone’s financial model.

5. The counterfactual that actually works

What if the same $108 billion (Goldman’s base-case value) were spent differently?

Imagine allocating roughly $40 billion to extend fiber backbones across underserved regions, $30 billion to expand community-owned wireless networks and reform spectrum policy so that local operators can thrive, $20 billion to provide affordable devices and digital literacy programs, and the remaining $18 billion to research and scale appropriate technologies such as TV White Space and mesh networks. That mix would connect billions rather than millions, create infrastructure that lasts for 20–30 years instead of five, and build domestic capacity instead of orbital dependence. It’s the kind of investment portfolio that multiplies inclusion instead of merely broadcasting connectivity.

That mix would connect billions, not millions, with infrastructure that lasts 20–30 years — and would build local capacity instead of orbital dependence.

6. The real development strategy

LEO satellites do have genuine roles: emergency response, maritime and aviation connectivity, remote research, defense. But they should supplement, not set, development strategy.

Public finance should go to ground networks, not space races. Let private capital chase orbital dreams for high-value users. Governments and development banks should invest where returns compound: open fiber, community networks, spectrum reform, digital literacy.

7. The real bottleneck is not in the sky

The limiting factor isn’t how many satellites we can launch — it’s how effectively we can use the data we already get. Earth observation, not internet beaming, will yield the next productivity gains if countries build institutions to act on that data.

The policy bottom line

Connecting the next billion through LEO could cost $50–100 per person in infrastructure; terrestrial networks can do it for $5–15. That’s the difference between reaching 100 million people and 1 billion with the same budget.

Connectivity isn’t a problem of altitude. It’s a problem of economics, governance, and trust.

The next digital revolution won’t be won in orbit — it will be wired, towered, and rooted firmly on the ground.