SpaceX was founded in March 2002. In roughly 25 years, it helped turn reusable rockets from an idea many considered fantasy into the operating model reshaping the launch industry.

Now look forward another 25 years, to roughly 2050, and consider a thought experiment: What if launching a kilogram to orbit costs about $100 by then?

That number is not a prediction. But it is no longer pure science fiction.

A recent study in PNAS Nexus assembled data on more than 4,400 launches from 1960 through 2025. In constant 2024 dollars, the researchers calculated that the average cost of sending a kilogram to low Earth orbit fell from more than $87,000 in 1960 to less than $4,000 in 2025.

They found that average cost declined 21.2 percent with each doubling of cumulative payload. Their central scenario projects approximately $1,600 per kilogram in 2030 and $300 in 2040. Extending that trajectory toward mid-century gets into the neighborhood of $100. (Read the study)

Experience curves are empirical relationships, not physical laws. Demand, competition, regulation, geopolitics, and orbital congestion could all change the trajectory. The study also notes that SpaceX currently carries roughly 75 percent of global payload, making its ability to scale a near-term concentration risk, even if reusable-launch technology and its associated learning diffuse more broadly over the next quarter century.

And an operator's cost is not the price quoted to a customer. If capacity remains constrained and competition limited, providers may retain much of the cost reduction as margin.

But the architectural effects can begin before $100 per kilogram appears on a public price sheet. A vertically integrated operator can design around its own cost base even while charging outside customers substantially more. If capacity expands and credible competition emerges, more of those savings should spread across the market.

So $100 per kilogram is best understood as a boundary condition worth testing: What changes when the underlying cost of transporting mass to orbit is no longer the constraint around which everything else must be organized?

From a national project to a freight bill

Start with a laptop weighing approximately two kilograms.

At the study's calculated 1960 industry average, its mass-equivalent transportation cost would have exceeded $170,000 in today's dollars. At the 2025 average, it is less than $8,000. At the study's 2040 central estimate, it falls to approximately $600. At $100 per kilogram, it is $200.

Of course, no one can walk up to a rocket with a laptop and purchase two kilograms of unused capacity. Orbit, volume, integration, safety, schedule, and minimum bookings all matter.

The distinction between cost and price is visible today. In 2019, SpaceX advertised a rideshare price of $1 million for up to 200 kilograms, or effectively $5,000 per kilogram for a full booking. Today, its rideshare offering starts at $350,000 for up to 50 kilograms, with additional mass priced at $7,000 per kilogram. The minimum ticket has fallen even as the nominal unit price has increased. Price reflects product structure, inflation, capacity, competition, and customer value, and not simply underlying cost. (Current SpaceX pricing; contemporaneous 2019 pricing)

Of course, published prices do not guarantee available capacity. As of this writing, industry reporting indicates that SpaceX has stopped accepting new Transporter and Bandwagon bookings beyond late 2028 because its Falcon 9 manifest is effectively full. (AIAA summary)

Something that once carried a six-figure transportation burden is beginning to look like premium freight, and could eventually look more like ordinary freight. But how quickly that happens for outside customers depends on capacity and competition as well as technology.

Now consider something much larger. The International Space Station has a mass of approximately 420 metric tons. It is longer than a football field, contains more living and working space than a six-bedroom house, and was assembled through 42 major flights, including 37 Space Shuttle missions. (NASA station facts)

A NASA analysis estimated the Space Shuttle's average full cost at approximately $1.5 billion per flight, or about $54,500 per kilogram at maximum payload. At that rate, 420 metric tons represents a mass-equivalent cost of approximately $23 billion. (NASA analysis)

At the 2025 industry average, the same mass represents roughly $1.6 billion. At the study's 2040 central estimate, less than $130 million. At $100 per kilogram: approximately $42 million.

That does not mean anyone could build another ISS for $42 million. It only gets the pieces into orbit. Modules, power, thermal management, docking systems, life support, software, integration, assembly, operations, and engineering would still dominate the undertaking. But it points toward a world in which transporting an ISS-scale mass costs tens of millions rather than billions.

Human spaceflight makes the distinction even clearer. In 2024, Axiom Space described the price of a roughly 10-day orbital mission as being in the mid-$60-million range per seat. (Axiom Space)

Now focus only on mass. An 80-kilogram passenger, a laptop, and roughly a month's open-loop supply of food, water, and oxygen amount to something on the order of 200 kilograms. The mass-equivalent transportation cost is approximately $800,000 at the 2025 industry average and about $20,000 at $100 per kilogram.

A month in orbit would not become a $20,000 vacation. People need a human-rated spacecraft, abort capability, life support, training, insurance, operations, a destination, and a safe trip home. Many of those costs scale with each passenger or each mission rather than with raw payload mass and, therefore, will not decline in lockstep with launch cost per kilogram.

The enormous gap between the cost of transporting the mass and the price of transporting the person shows how much of the system lies beyond the rocket's raw payload cost.

In short, launch doesn't become free at $100/kg. But it does become small enough that you have to reconsider almost every other part of the system.

The obvious calculation is the least interesting one

The obvious way to discuss cheaper launch is to take an existing satellite, multiply its mass by a lower number, and calculate the savings.

But that is like imagining the internet as a cheaper fax machine. It misses the change in the system.

Space hardware is extraordinarily expensive partly because it is designed around the assumption that every kilogram is precious and every mission may be the only chance to get it right. Engineers use exotic materials, custom components, complex folding mechanisms, extensive testing, and painstaking mass reduction. They build for years of unattended operation because repair or replacement may be impossible.

If launch becomes inexpensive, frequent, and dependable, designers get different options.

They may use heavier but less expensive materials. They can add shielding, structural margin, redundancy, consumables, and spare parts. They can adopt standardized components and simpler manufacturing methods. They can build systems that are upgraded every few years instead of systems expected to remain technologically frozen for decades.

The biggest cost reduction may not appear in the rocket bill. It may appear in the cost of everything riding on the rocket.

The question shifts from “How can this spacecraft survive forever?” to “How can this system be made easy to build, launch, operate, refresh, and retire, again and again?”

What happens to manufacturing?

In 2017, the IDA Science and Technology Policy Institute examined possible markets for a privately owned and operated space station. Its analysis assumed a price of approximately $20,000 per kilogram to transport encapsulated cargo to orbit. It considered 21 potential revenue-producing activities, including silicon-carbide wafers, exotic optical fiber, on-orbit assembly, research, tourism, and media. (Read IDA Paper P-8247)

The study was hardly promotional. It concluded that only its high-revenue, low-cost scenario produced an attractive business case. It also emphasized how uncertain the projected revenues remained.

Revisit some of those cases at $100 per kilogram, however, and the thresholds change radically.

Take silicon-carbide wafers, which the study examined as candidates for microgravity processing to remove defects. At its assumed cargo price, transporting one 30-gram wafer to a station costs approximately $600. At $100 per kilogram, the outbound transportation cost falls to $3.

IDA also found that parabolic aircraft flights could perform the process for roughly $217 per wafer, far less than the estimated $600 orbital transportation charge. At $3 of outbound launch cost, that comparison deserves to be run again.

But cheap launch does not prove the orbital case. The decisive questions become whether microgravity produces a large and repeatable quality advantage, how many usable wafers emerge, what station time and equipment cost, what it takes to bring the product home, and whether customers will repeatedly pay enough for the result.

The same discipline applies to ZBLAN optical fiber. IDA cited then-current prices of $175 to $1,000 per meter and estimated that one kilogram of ZBLAN could yield 2.2 kilometers of fiber, implying potential product value between $385,000 and $2.2 million per kilogram. Against those assumed values, a $100 outbound launch charge is irrelevant.

But those calculations were not proof that large quantities of microgravity-produced ZBLAN would achieve the assumed quality, find buyers, or command the same prices. Production consistency, throughput, return logistics, terrestrial alternatives, and demonstrated willingness to pay still determine whether a market exists.

Lower launch cost expands the set of technically plausible businesses. It does not establish demand.

A market still requires an identifiable buyer, a product that delivers measurable value, a price the buyer will repeatedly pay, and an operating system capable of delivering it reliably.

Cheap launch may even reverse some seemingly futuristic conclusions. Printing a routine replacement part aboard a station sounds advanced. But if terrestrial factories can produce standardized parts cheaply and frequent flights can keep orbital inventory stocked, shipping the part may be easier than maintaining a general-purpose factory in orbit.

The more compelling in-space manufacturing opportunity may instead be building things that cannot be launched in one piece: very large antennas, telescopes, solar arrays, radiators, habitats, and other structures that no longer need to fit inside a rocket fairing or survive launch as completed systems.

Imagine assembling a large telescope aperture or power array from standardized truss and panel units delivered across multiple flights, rather than designing a single origami-like structure around one rocket fairing.

Manufacturing processes have already been demonstrated in space at experimental scale. The relevant question is which products gain a decisive advantage once launch cost is no longer the first objection.

What would an orbital data center look like?

Space-based data centers are a useful second example because they expose the difference between cheaper transportation and a genuinely different architecture.

The obstacles are real. Computing consumes enormous power and produces enormous heat. Space may be cold, but there is no air to carry heat away. Large radiators are required. Radiation damages electronics. Communications, maintenance, orbital debris, equipment disposal, and latency all matter.

Cheaper launch does not repeal physics.

It does, however, change how systems can be built around the physics.

The instinct today is often to imagine an orbital data center as one fully integrated spacecraft. Terrestrial data centers do not work that way. The building, electrical service, cooling system, communications, and security infrastructure last much longer than the servers. Compute and storage are replaced continuously as technology improves or equipment fails.

An orbital data center could evolve along similar lines. Solar arrays, radiators, communications, propulsion, shielding, and robotic handling become long-lived capital infrastructure. Compute and storage arrive in standardized modules. New modules are installed; obsolete or failed modules are removed. Spare capacity is kept on hand. Customers buy power, cooling, connectivity, and computing capacity rather than an entire bespoke spacecraft.

Versions of this idea are already being proposed, but the architecture is not settled. Perhaps the winning model is a shared platform with replaceable compute containers. Perhaps it is a set of integrated tiles, each with its own computing, power, and heat rejection. Perhaps it is a distributed constellation rather than a single orbital campus.

None of those models has yet proved it can beat terrestrial alternatives.

At $100/kg, the winning design may not be the lightest or the longest-lived. It may be the one that is easiest to manufacture, expand, upgrade, repair, and replace.

In other words, orbital infrastructure would start to look more like infrastructure everywhere else.

The Next Space Economy is built on repetition

The phrase “space economy” invites grand forecasts: trillions of dollars, lunar settlements, orbital factories, space hotels, and industries that do not yet exist.

Some may become large markets. Others may remain demonstrations, government programs, or compelling ideas without enough demand to support them. Lower launch costs alone will not tell which is which.

The more useful test is whether an activity can become repeatable.

• Can hardware be produced in families rather than as one-off masterpieces?

• Can components connect through common interfaces?

• Can capacity be added in useful increments?

• Can customers buy a recurring service instead of funding a unique mission?

• Can equipment be replenished, upgraded, and retired predictably?

• Can insurers, lenders, investors, and regulators evaluate a category of activity rather than start from zero every time?

A voyage is a mission. A shipping lane is a market.

These are not supporting details. They are how markets form.

For companies, the opportunity is to design for repeatability before the market demands it. For investors, it is to identify businesses that become stronger as launch gets cheaper rather than businesses whose advantage depends on scarcity or exquisite mass optimization. For policymakers, it is to move from approving singular missions toward governing recurring operations, fleets, hardware families, orbital traffic, and responsible retirement.

There will be costs omitted from the optimistic scenarios. More hardware in orbit creates a greater need for tracking, maneuvering, deorbiting, recycling, and managing reentry. Cheap launch may lower the cost of putting things up while increasing the importance of bringing them down safely.

As the economic value of orbital assets grows, coordinated traffic and debris management will become economic infrastructure rather than peripheral policy concerns. Recurring orbital commerce will be difficult to finance if collision risk, operating priority, liability, and end-of-life obligations remain unclear.

And then there are the questions that capture the imagination. What does this do to the cost and design of a commercial space station? How much could research time in orbit fall? When does a week at a low-Earth-orbit destination become a luxury purchase rather than an ultra-wealthy adventure? What happens to the cost of building and supplying systems around the Moon?

The $100-per-kilogram question is not really about the rocket. It is about what companies, investors, and governments do when mass is no longer the constraint around which everything else must be organized.

The winners in the Next Space Economy may not be those that build the most exquisite spacecraft. They may be those that learn how to build, operate, replenish, and retire space systems repeatedly.

Orbital Progress studies how technical possibilities become repeatable, financeable markets, and what companies, investors, and policymakers can do to accelerate that transition without mistaking activity for demand.