Amazon is joining a long list of companies planning constellations consisting of hundreds and thousands of satellites with the stated aim of providing Internet connectivity to the world’s 3.8 billion without access to the Internet.

The company recently filed an application with the International Telecommunication Union (ITU) for a constellation of 3,236 satellites arranged in three orbits. There will be 784 satellites in a 590-kilometer orbit, 1,156 in a 630-kilometer orbit, and 1,296 in a 610-kilometer orbit. Known as Project Kuiper, this Non-Geostationary Orbit Satellite (NGOS) constellation will provide low-latency, high-speed broadband connectivity to unserved and underserved communities around the world.

To be sure, it is already a crowded market that Amazon is entering. Some of the world’s best-known companies have been working on similar plans for years. Last year, SpaceX launched the first two prototype satellites for its Starlink broadband data constellation, which ultimately will consist of nearly 12,000 satellites in low-Earth orbit (LEO). SpaceX’s billionaire founder, Elon Musk, has said he expects revenue from the Starlink’s internet service “to help fund his vision of building a city on Mars!”

Similarly, there is OneWeb which is backed by Airbus, Softbank, and others. The company already launched its first six broadband satellites in February and plans to put hundreds more in orbit over the next year or two. Last month, the consortium announced a US$1.25 billion funding round to support the ramp-up in operations.

Meanwhile, Canadian satellite company Telesat launched its first prototype broadband satellite into LEO last year. It plans to launch hundreds more to provide first-generation broadband services in the early 2020s. Boeing, Facebook, and Luxembourg-based LeoSat also have laid out similar plans for space-based Internet access. (Table 1).

*Initial constellation

Table 1: NGOS Satellite Constellations

Although several companies are chalking out plans for broadband delivery via NGOS, launching a satellite constellation with hundreds of satellites is fraught with a host of technical and commercial risks. Licensing the constellations in every country around the world, technical challenges such as successfully launching hundreds of satellites, operating in new microwave frequency bands, dealing with issues of interference with other satellites as well as ground-based services are only some of the hurdles. Besides, the new NGOS operators will face competition from existing Geostationary Earth Orbit (GEO) based services already operating in their target markets.

Given the multi-billion-dollar costs of building and replenishing these constellations, the licensing difficulties, intense competition from GEOs, etc., Counterpoint Research believes that most, if not all, of these NGOS start-ups, will never become fully operational.

Indeed, this looks very much like a re-run of the 1990s NGOS boom in which 6-7 constellations were developed to provide mobile voice communications in underserved regions of the world. Back then, there was alsoTeledesic, a 900-satellite plan backed by Craig McCaw and Bill Gates, with the same laudable idea of providing Internet access to the world’s poorest communities. It also never got off the drawing board.

Bezos’s Kuiper constellation, however, may be an exception. The Amazon boss is no stranger to the space business. He founded a space vehicle launch company called Blue Origin more than 18 years ago and is rumored to be funding the company at a rate of US$1 billion per year. The company is developing a reusable rocket called New Glenn, and its first flight could be in 2021. It could be used to launch Kuiper’s satellites.

Clearly, Bezos is committed to the space business, and with Amazon’s market capitalization currently close to US$900 billion, he certainly could afford to fund Project Kuiper. However, he will have to come up with a much better business plan than providing Internet connectivity to the world’s poorest if Kuiper is ever to have any chance of becoming profitable.