The New Infrastructure

We have been watching rockets for spectacle. The launches, the landings, the failures that bloom into fireballs against a dark sky, these are the images that dominate coverage and capture attention. But the consequential story is not the vehicles. It is the payload.

Thousands of satellites are being placed into low Earth orbit, and they are doing something that changes the map: they are turning connectivity into a utility that does not depend on terrestrial infrastructure. Not as a concept. Not as a pilot program. As a commercial service, at scale, serving millions of paying customers across more than a hundred countries.

This is not a technology demonstration. It is an infrastructure shift, the kind that, historically, has reshaped economies and redrawn the boundaries of what is possible. The pattern is familiar. The scale is not.

The economics that changed everything

Infrastructure revolutions begin with cost collapse. Not with invention, not with vision, with the moment when deploying a thing becomes cheap enough to deploy it everywhere. The question is never “can we build it?” but “can we afford to build enough of it?”

The Space Shuttle carried payload to orbit at approximately $54,500 per kilogram. SpaceX’s Falcon 9 does the same work for roughly $2,700 per kilogram, a twentyfold reduction that has already been realized, not projected. SpaceX’s internal marginal cost is estimated closer to $629 per kilogram. And Starship, if it achieves full reusability, targets $67 to $100 per kilogram in the near term, with a theoretical floor around $10 per kilogram. That would represent a further thirty- to fiftyfold drop from today’s commercial rate.

These are not incremental improvements. They are the kind of cost discontinuities that precede every major infrastructure expansion in modern history. Undersea telegraph cables became ubiquitous not when the technology matured, it was mature by the 1870s, but when the cost per message dropped below the threshold at which businesses could justify using them routinely. Cell towers proliferated not when the engineering was proven but when deployment costs fell enough to justify rural coverage, transforming mobile telephony from an urban luxury into a universal assumption.

The same dynamic is now playing out in orbit. SpaceX conducted 165 Falcon 9 launches in 2025, one every 2.2 days. That cadence is not experimentation. That is industrial-scale deployment, governed by the same logic that drove the build-out of every previous network: the economics work, so you build as fast as the economics allow.

It is worth pausing on what a twentyfold cost reduction means in practice. It does not mean you do the same thing for less money. It means you do things that were previously irrational. A constellation of 7,500 satellites was economically absurd at Shuttle-era prices. At Falcon 9 prices, it is a business. At Starship prices, it becomes a commodity. Each step down the cost curve does not merely improve the existing market, it creates a new one. This is the lesson of every infrastructure cost collapse, from steel rails to fiber optics: the demand that matters most is the demand that does not yet exist because the price has not yet fallen far enough to call it into being.

The constellations

What this economics enables is the mega-constellation, a design philosophy that inverts six decades of orbital communication strategy.

The traditional approach placed a small number of large, expensive satellites in geostationary orbit, 36,000 kilometers above the equator. Each one cost hundreds of millions of dollars, took years to build, and served as a single point of failure for the coverage area it served. The new approach places thousands of small, relatively inexpensive satellites in low Earth orbit, between 340 and 550 kilometers up. Lower altitude means lower latency. Larger numbers mean global coverage and graceful degradation, lose a satellite, and the network routes around it.

Starlink is the most advanced of these constellations: more than 7,500 satellites in orbit, serving over 10 million subscribers across 114 countries. It is operating at the scale of a major telecommunications provider, not a technology experiment. Amazon’s Project Kuiper, now branded Amazon Leo, has launched over 200 satellites and is targeting commercial service in five countries by the first quarter of 2026. Eutelsat OneWeb operates a constellation of more than 600 satellites and has placed a $2.56 billion order with Airbus for 440 next-generation replacements.

And then there is the strategic dimension. China is building parallel infrastructure on a scale that signals how seriously nation-states take orbital communications as a matter of sovereignty. The Guowang constellation plans approximately 13,000 satellites, with 113 already in orbit. Qianfan, sometimes called China’s Starlink, has filed for more than 15,000. Combined, these programs represent nearly 28,000 planned satellites from a single nation, built and operated independently of Western systems.

Multiple countries are investing as if space-based communication is strategic national infrastructure, because it is. The logic is the same logic that drove governments to build national telephone networks and subsidize broadband deployment: connectivity is a precondition for economic participation, and dependence on another nation’s connectivity is a strategic vulnerability.

The shift from a few expensive geostationary platforms to thousands of cheap low-orbit ones is the shift from mainframes to distributed computing, applied to the orbital domain. A GEO satellite is a mainframe, powerful, centralized, and catastrophic to lose. A LEO constellation is a distributed system, individually expendable, collectively resilient, and improvable through continuous deployment rather than generational replacement cycles.

Direct-to-device: the invisible threshold

For all the progress of the mega-constellations, there was a bottleneck that kept space-based communication from becoming true infrastructure: the ground terminal. Starlink required a dish, a dedicated piece of hardware that had to be purchased, installed, and powered. This made it a service for fixed locations and specialized vehicles, not for the general population walking around with phones in their pockets.

That bottleneck is dissolving.

T-Mobile and SpaceX launched commercial direct-to-cell service in July 2025. The system delivers text messages and images to unmodified smartphones, no special hardware, no app, no configuration. More than 650 direct-to-cell satellites are in orbit, and the service has reached over 12 million customers. It is not a prototype. It is a product.

AST SpaceMobile launched its BlueBird 6 satellite in December 2025, the largest commercial communications array ever deployed in low Earth orbit. It delivers broadband speeds of up to 120 megabits per second to standard mobile phones, using the same cellular frequencies those phones already support. The satellite is, in effect, a cell tower that happens to be in space.

And Apple, with characteristic quietness, put satellite communication hardware into every iPhone starting with the iPhone 14 in 2022. Three years later, hundreds of millions of satellite-capable devices are in pockets around the world. The initial use case was emergency SOS. The trajectory points toward routine connectivity.

This convergence, dedicated constellations, carrier partnerships, and consumer device integration, is not three separate trends. It is one trend, approaching from different directions. SpaceX is working downward from orbit to the handset. Apple is working upward from the handset to orbit. The carriers are working outward from their existing networks to fill the gaps. The result is the same: the elimination of the ground terminal as a barrier to satellite communication.

This is the threshold moment. When the phone in your pocket connects to a satellite without you knowing it, when the signal arrives from orbit as seamlessly as it arrives from the tower down the street, the infrastructure has become invisible. And invisible infrastructure is the definition of successful infrastructure. You do not think about the undersea cable when you load a website served from another continent. You do not think about the cell tower when you make a call. The infrastructure has succeeded precisely when you stop noticing it.

The infrastructure argument

The pattern repeats with remarkable consistency.

Undersea telegraph cables in the 1860s were expensive curiosities, the first transatlantic cable of 1858 failed within weeks, and its replacement took eight years and several fortunes to complete. By the 1880s, the cable network was the nervous system of global commerce, transmitting commodity prices, diplomatic messages, and financial transactions at a speed that reorganized markets. By the early twentieth century, no serious commercial enterprise could operate without access to the cable network. The infrastructure had become a precondition for participation in the global economy.

Cell towers followed the same arc. In the early 1990s, mobile phones were conspicuous and unreliable, dropped calls, dead zones, devices the size of bricks. By the 2000s, the tower network had become so dense and capable that wireless connectivity stopped being a feature and started being an assumption. Entire industries, ride-sharing, mobile banking, social media, exist because someone built enough towers in enough places to make wireless connectivity ambient and reliable.

Space-based communication is following this trajectory. The question is no longer “can we deliver broadband from orbit?”, Starlink has answered that with 10 million subscribers. The question is “what happens when every point on Earth has connectivity?” Not most points. Every point. Oceans, polar regions, deep rural areas, disaster zones, airspace, shipping lanes, the places where ground infrastructure is either uneconomical or impossible.

The second-order effects dwarf the primary use case. Maritime shipping gains continuous, high-bandwidth connectivity for fleet management and autonomous operations. Aviation gets reliable passenger internet and real-time telemetry. Remote communities access education, healthcare, and financial services without waiting for someone to run fiber. Disaster response operates when terrestrial networks are destroyed. Military communications become resilient against the denial of ground infrastructure. The Internet of Things extends to places where no one will ever build a cell tower.

The satellite services market is projected to exceed $300 billion by 2030, a figure comparable to submarine cable and cell tower infrastructure combined. This is not a niche technology serving an edge case. It is a new layer of global infrastructure, and it is being built at a pace that outstrips every terrestrial precedent.

The unresolved risk

Every infrastructure has its systemic risk, the vulnerability that, if left unmanaged, threatens the viability of the entire system.

For undersea cables, it was geopolitics and physical vulnerability. Cables were, and still are, cut by anchors, earthquakes, and occasionally by deliberate sabotage. The response was redundancy: enough cables on enough routes that the loss of any one was an inconvenience, not a catastrophe.

For cell towers, it was spectrum scarcity. The radio frequencies that make mobile communication possible are finite, and the process of allocating them among competing users became one of the defining regulatory challenges of the late twentieth century. The response was a combination of regulation, technology, and market mechanisms, spectrum auctions, frequency reuse, ever-more-efficient encoding.

For orbital infrastructure, the systemic risk is debris.

There are currently more than 54,000 tracked objects larger than 10 centimeters in orbit. An estimated 140 million fragments smaller than one centimeter travel at velocities sufficient to damage or destroy operational satellites. The planned mega-constellations would add more than 65,000 satellites to an already crowded environment, and each satellite that fails or is retired becomes, if not properly disposed of, another piece of debris.

The Kessler syndrome, a theoretical cascade in which collisions generate debris that causes further collisions, progressively rendering entire orbital bands unusable, is the existential risk of this infrastructure layer. It is not a near-term crisis, but it is the kind of slow-building, hard-to-reverse problem that has historically caught infrastructure planners off guard.

The regulatory response is beginning to take shape. The FCC’s five-year deorbit rule, effective since 2024, is the first serious framework treating orbital space as a managed commons rather than a frontier. ESA’s ClearSpace-1 debris removal mission is planned for 2026. But regulation lags deployment by years, as it always does with new infrastructure. The fiber-optic cables were laid before anyone thought seriously about submarine cable security. The towers were built before spectrum allocation caught up with demand.

The debris question is this generation’s spectrum allocation problem, solvable in principle, catastrophic if ignored, and currently being addressed at a pace that does not match the speed of deployment.

What the infrastructure makes possible

The most consequential infrastructure shifts are the ones that happen before most people notice. The internet was “just email” before it remade commerce. Mobile was “just phone calls” before it remade computing. Each time, the infrastructure was dismissed as a niche technology serving a narrow use case, right up until it became the platform on which entire industries were built.

Space-based communication is in the same early-but-already-irreversible phase. Ten million subscribers. More than 7,500 operational satellites. Direct connection to unmodified phones. Multiple nations building redundant constellations on the assumption that orbital connectivity is a strategic necessity. A cost curve that is dropping faster than any terrestrial equivalent in history.

The signs of irreversibility are not in the technology announcements. They are in the capital commitments. Amazon is spending more than $10 billion on Project Kuiper. SpaceX has raised tens of billions to build Starlink and Starship. China is committing state resources to Guowang and Qianfan at a scale that only makes sense if the objective is permanent infrastructure, not a technology experiment. When this much capital flows in this many directions toward the same outcome, the outcome is no longer speculative. The debate has moved from “whether” to “how fast” and “controlled by whom.”

The infrastructure is being built now, at industrial scale, by entities that are spending tens of billions of dollars on the bet that ubiquitous connectivity from orbit will be as fundamental to the next era as undersea cables were to global trade and cell towers were to mobile computing.

The revolution is not the rockets. The rockets are the means. The revolution is what they are putting into orbit, and what that connectivity will make possible once it becomes, as all good infrastructure eventually does, invisible.