The Future Is Built

We build the future we read about as kids.

In October 1945 the radio magazine Wireless World published a technical article by Arthur C. Clarke, who had joined the British Interplanetary Society in 1934, served in World War II as a radar technician and instructor in the Royal Air Force, and would become one of science fiction's best-known writers. Three relay stations, spaced 120 degrees apart in an orbit about 35,786 kilometers up, he argued, could carry radio and television to the whole planet. The stations in his article had crews, living quarters and laboratories, and mirrors to gather sunlight for power. [1]

The author read Clarke and he strongly believes that those that invented the space industry did as well.

Twelve years later, on January 31, 1958, the United States launched its first satellite, Explorer 1. The Army had modified a missile into the Jupiter-C rocket, and the Jet Propulsion Laboratory, a Caltech-run lab that would soon join NASA, prepared its three upper stages and built the satellite, all in 84 days. It was the first satellite to detect the Van Allen radiation belt. [2] That belt is a zone around Earth where fast-moving charged particles are trapped, and one of them is enough to flip a bit in a computer chip. A steady dose wears the chip out. So chips made for space are "radiation-hardened": built on special materials, with extra copies of the circuitry and memory that checks and corrects itself, all of which makes them slower and more expensive. [3]

On July 10, 1962, four years later, a company did what only governments had done. AT&T had built Telstar 1 and spent about $50 million on it, and it paid NASA $3 million a launch. Telstar was the first privately sponsored space launch, and it relayed the first live television and telephone calls across the Atlantic. [4] John Pierce of Bell Labs had named the transistor in 1948 and wrote science fiction under the pen name J. J. Coupling; he was a leading figure in Telstar, and Clarke later credited him, with Harold Rosen, as a father of the communications satellite. [5]

Clarke never built a satellite. Harold Rosen did the engineering, at Hughes Aircraft, with Don Williams, Tom Hudspeth and John Mendel. What he saw was that a satellite spun like a top, nudged by thruster pulses timed to the spin, could be steered by a simple control system that would last, and could carry light electronics. His management was reluctant. He persuaded it to pay for a prototype first, and then persuaded the government to fund the flight program: in 1961 NASA paid Hughes $4 million to build Syncom. Syncom 2, launched on July 26, 1963 and weighing about 78 pounds, became the first operational geosynchronous satellite, which is to say one that circles the Earth once a day and so keeps to roughly one spot in the sky. Syncom 3 carried pictures of the 1964 Tokyo Olympics. Rosen received the National Medal of Technology in 1985. [6] In my reading, what Clarke handed them was a vision, not a design: the readers who built it were the ones who recognized the moment the technology caught up.

But Telstar was a private satellite on a government rocket, and for the next quarter century that pattern stayed true: the shuttle had been meant to fly the nation's payloads, commercial ones included. On August 15, 1986, after the loss of the space shuttle Challenger, President Reagan's launch strategy barred NASA from flying commercial satellites on the shuttle, unless the shuttle was the only vehicle that could carry them or a foreign-policy or national-security interest required it. The aim, as reported then, was to stop the shuttle from competing with private launch companies. The Commercial Space Launch Act of 1984 had already set up licensing for private launches. [7] A private satellite and a commercial launch are different things, and the second took decades longer than the first.

Government rockets were built by contractors, and there were few of them. Boeing and Lockheed Martin merged their launch businesses into United Launch Alliance on December 1, 2006, combining the Atlas and Delta rockets; its main customers are the Defense Department and NASA. [8] NASA's own heavy rocket comes out of the same world. The NASA Authorization Act of October 11, 2010 cancelled the Constellation program, and its Ares rockets were redeveloped into the Space Launch System. [9] Boeing, Northrop Grumman and Aerojet held 71% of the system's contract funding in 2019, on cost-plus contracts, which pay a contractor's costs and then a fee on top. In 2022 NASA's inspector general put the production and operation of the rocket and the Orion capsule at $4.1 billion a launch for the first four Artemis missions, and added that cost-plus contracts had worked to the contractors' advantage. [10] The break came in August 2006, when NASA paid SpaceX $278 million and Rocketplane Kistler $207 million under a new kind of funded agreement, each had to raise private money alongside the contracts and fly cargo to orbit. Rocketplane Kistler's agreement was ended when it could not raise the financing. SpaceX's was not. [11]

What changed after that was the price. A 2018 NASA Ames paper compared a government launch with a commercial one: the shuttle at about $1.5 billion a flight, which is $54,500 a kilogram to low Earth orbit (the first few hundred miles up), and Falcon 9 at an advertised price of $62 million, $2,720 a kilogram, roughly twenty times less. [12] Price is not cost, though. The first number is what a launch cost the government, the second is what a company charges, and what it costs the company is not public. The phone arrived on the same curve. NASA launched three PhoneSats on April 21, 2013, built from off-the-shelf phones, the first pair around an HTC Nexus One, for about $3,500 in parts each. [13]

The payload changed with the vehicle. When a rocket flew once, the satellite and the rocket were usually designed together, as Explorer 1 and the Jupiter-C had been. [2] A standard small-satellite size, the CubeSat, had existed since 1999, with a standard tube to release it, [14] and PhoneSat was built to that size. [13] On January 24, 2021, SpaceX's Transporter-1 turned the standard into a product, selling space on one reusable rocket at a base price of $1 million for up to 200 kilograms. [15] In my reading, reusing the rocket, and releasing cargo the same way every time, pushed payloads toward standard sizes.

Who captured the savings is a matter of public record. SpaceX priced the largest initial public offering in history on June 11, 2026, at $135 a share, raising about $75 billion at a valuation of about $1.77 trillion. [16] Jeff Bezos took the other route: in April 2017 he said he would sell about $1 billion of Amazon stock a year to pay for Blue Origin. [17] By Forbes' real-time list, Musk stood at roughly $0.9 to $1 trillion in mid-2026 and Bezos at about $250 billion. [18] In dollar terms Norway's 2025 GDP, about $531 billion, is just over half of Musk's fortune, and Bezos's fortune is about the size of Hungary's economy, roughly $247 billion, a line about 60 countries fall below, though a fortune is a stock and GDP is a year's output. [19] Musk had started SpaceX and Tesla with money he made at PayPal. [20] In my reading, one company started on earlier money and then built the cost curve and went public on it. The other was to be funded out of a fortune made somewhere else.

Private does not mean unfunded. About a fifth of SpaceX's 2025 revenue came from the US government, and the company launched 11 of the 12 medium- and heavy-lift national security missions that year. [21] It also had to raise money from outside. Musk's own $100 million went into the first launches, three Falcon 1 flights failed, and the fourth reached orbit on September 28, 2008, with fresh investment closing weeks before; within months NASA awarded the company a $1.6 billion contract for 12 cargo flights to the space station. [22] NASA awarded commercial crew contracts worth $6.8 billion to Boeing and SpaceX on September 16, 2014, and in May 2023 chose a Blue Origin team for a $3.4 billion lunar lander contract. [23] It has also paid commercial station developers: $217.5 million to Voyager Space's Starlab and $172 million to Blue Origin's Orbital Reef. [24] The government that once launched Telstar for AT&T now buys rides, landers and stations from the companies it helped build. In my reading, that puts the taxpayer on both sides of the bill: customer of the less expensive launch, and operator of the station that has to dodge what the less expensive launch put in orbit.

The next generation grew out of that same drop in cost. Planet Labs was founded by three NASA researchers, Will Marshall, Robbie Schingler and Chris Boshuizen, and Marshall and Boshuizen had led the NASA team whose PhoneSats reached orbit in 2013 at $7,000 each. [25] Planet went public on the New York Stock Exchange in December 2021. [26] Rocket Lab, whose second test flight in January 2018 carried a Planet satellite, took a different path. [27]

Rocket Lab's early money came from venture firms, Khosla and Bessemer, and from Lockheed Martin, itself a major government contractor. [28] After Rocket Lab listed in 2021, Khosla Ventures held about 26 percent of its voting power and Bessemer about 18 percent. [29] In-Q-Tel, the venture firm that invests for the US intelligence community, also lists Rocket Lab among its portfolio companies, now as an exit. [30] In 2025 Rocket Lab reported $602 million in revenue and a backlog of $1.85 billion that included an $816 million Space Development Agency contract, its largest. [31] In my reading, size opens up options. A small rocket has more places to launch from. Electron carries up to 300 kilograms to low Earth orbit, and Rocket Lab flies it from Mahia in New Zealand and from Wallops Island in Virginia, where NASA has a flight facility. [32]

Others follow the same shape. Firefly's Blue Ghost touched down on the Moon in March 2025 on a NASA contract, and Firefly went public that August. [33] Intuitive Machines' Odysseus touched down in February 2024 on a NASA fixed-price contract of a little over $100 million. [34] And Eric Schmidt, the former Google chief executive, took a controlling stake in Relativity Space in March 2025, according to reports. [35] In my reading, none of them has left the government behind. Rocket Lab's largest contract is a government one, and the Firefly and Intuitive Machines landers flew for NASA. The bank that lends to these companies says venture money and defense demand now reinforce each other. [36]

Lower cost changes which parts are worth buying. Hardened chips need their own kind of quality assurance, years of testing to prove a part survives, so their designs trail the ones in a phone. [3] The RAD750 processor is based on a commercial design from the 1990s, was released in 2001, first flew in 2005, runs at 110 to 200 megahertz, lists above $200,000, and flies on the Perseverance rover and the James Webb Space Telescope. [37] That premium pays where a failure cannot be repaired or replaced: a space telescope, a probe beyond Earth orbit. It stops paying where the satellite is expected to be replaced. SpaceX designs Starlink satellites for about five years and, between December 2025 and May 2026, retired 260 of them. [38] In my reading, that is an actuarial argument (the arithmetic insurers use), not an engineering one -- pay for hardening when the expected loss exceeds the premium, and falling launch cost shrinks the loss in low orbit.

Clarke's 1945 article says nothing of debris, insurers or a bill. [1] Then the argument runs into its limits. The first is fleet size. If each satellite had a one-in-a-million chance of causing a serious incident in a year (an illustration, not a measured rate), a fleet of 10 would carry about a 0.001% chance and a fleet of 100,000 about 9.5%. The same small risk becomes close to certain somewhere in a large fleet, and one collision that makes debris is not an independent event. The second is disposal. The FCC adopted a rule on September 29, 2022, and applied it from September 29, 2024, requiring new low-orbit satellites to come down within five years of the end of their mission. [39]

The third is people. The International Space Station moves when the chance of collision passes 1 in 100,000 or 1 in 10,000, measured against a zone about 25 kilometers wide on each side, and it has done so at least 29 times since 1999. [40] Of roughly 14,500 to 18,500 active satellites in 2026, about 70% are Starlink, which can steer out of the way. [41] CubeSats are often built without propulsion. [42] The station carries the fuel, the schedule and the crew risk of dodging satellites that cannot dodge it.

It also takes cooperation across networks. SpaceX, NASA, the NRO and other countries' agencies each run their own ground stations, and a low-orbit satellite is in range of one for only about 5 to 12 minutes a pass. [43] The shuttle shows what a crewed vehicle once got: each emergency landing site, in France, Spain, Morocco and The Gambia, had its own international agreement (France's was signed in 2005) and staff from NASA and the Defense Department on the ground. [44] Civil coordination of satellite traffic began in September 2024 with nine operators. [45]

In my reading, the less expensive satellite did not remove the cost of safety. It moved the cost onto whoever can least avoid it, which today is the crewed station. For an investor, ask who is paying to avoid the collision, and what happens to that bill when the fleet grows tenfold.

Minds are built by what they read, as How to Build a Mind argued. The future is what those minds build.

Sources

The launch-cost comparison sets what a launch cost the government against what a company advertises; SpaceX's own cost per launch is not public. The risk figure is an illustration. Satellite counts differ by tracker, and the share of CubeSats without propulsion and the 2024 coordination scale have not been checked against current data. The Starlink design life and retirement count are as reported. The reading that the cost has moved onto the crewed station, the actuarial framing and the standardization claim are the author's interpretation. This is not investment advice.