The Spectrum

Sixty-five dollars. That was the average monthly revenue per wireless subscriber in the United States in 2000 — the year before the iPhone existed, before mobile data was a meaningful category, before anyone used the phrase "dumb pipe." By 2020, the number was under $45. In the same period, data volume per subscriber grew by roughly a hundredfold. Revenue down 30%. Usage up 10,000%. The ratio of revenue to consumption is the cellular industry's version of the utility treadmill — except the treadmill in this case was built by competition, not regulation, and competition doesn't guarantee a floor.

Monopoly by Physics

Wireless introduced a variant of the natural monopoly problem that had no precedent in electricity or telephony. The radio spectrum is finite — there are only so many frequencies usable for wireless communications in any geographic area. This scarcity is physical, not economic. You can build a second power plant. You can dig a second trench. You cannot create additional radio spectrum. The government must allocate it, and allocation is inherently political.

The FCC's original cellular licensing in the early 1980s granted two licenses per geographic market: one to the incumbent wireline telephone company and one to a competitive applicant. Duopoly by design. Regulators recognized that spectrum scarcity precluded traditional competitive markets, but creating a pure monopoly — without the rate-of-return regulatory protections that governed wireline telephony — was politically impossible. Two providers it was.

The capital costs are enormous at every layer: spectrum licenses (now auctioned for tens of billions), radio towers every few miles, backhaul fiber connecting towers to the core network, core switching equipment. These fixed costs spread over subscribers create the familiar average-cost-decline-with-scale dynamic. But unlike wireline, where geographic monopoly is nearly absolute, wireless introduced spectrum as a tradeable asset — licenses could be resold, consolidated, accumulated. The FCC's first spectrum auction in 1994 began a process by which spectrum concentrated rapidly into a small number of operators with national coverage.

The Consolidation

By the mid-2000s, most original cellular licenses had consolidated into AT&T and Verizon. The mechanism was not just acquisition. It was spectrum quality.

AT&T and Verizon both held spectrum in the 850 MHz cellular band, which propagates over longer distances and penetrates buildings more effectively than the higher-frequency PCS spectrum (1900 MHz) that Sprint and T-Mobile held. This asymmetry meant AT&T and Verizon could achieve equivalent coverage with fewer towers — producing lower unit costs and justifying premium pricing. The FCC found that AT&T and Verizon experienced lower churn and charged more per subscriber — the classic indicators of durable market power rooted in infrastructure advantages.

Their rational bid for additional spectrum incorporated not only the spectrum's productive value but its foreclosure value: the strategic benefit of preventing rivals from closing the coverage gap. Buy spectrum you don't immediately need — and your competitor can't close the gap that makes your network more valuable. The result was a permanent oligopoly, produced not by regulatory grant (as in wireline telephony) but by the compounding advantages of early spectrum allocation, scale-driven cost advantages, and the network effect of national coverage.

The Dumb Pipe

The smartphone changed everything about wireless usage and nothing about wireless economics. After the iPhone launched in 2007, mobile data volume exploded. Consumers streamed video, downloaded apps, ran cloud services, consumed content at rates that would have been unimaginable to the engineers who designed 2G voice networks. The carriers had built the infrastructure that made this possible. The value accrued to someone else.

Revenue per bit fell faster than costs per bit, even as total data traffic grew exponentially. The carriers invested billions in spectrum and 4G LTE infrastructure. The returns accrued to Apple (which monetized the device), Google (which monetized the data through advertising), Netflix (which monetized the bandwidth through subscriptions), and every platform operator that rode the network without paying for it. This is the electric utility building a grid that primarily benefits appliance manufacturers — except the appliance manufacturers captured most of the economic surplus and the utility didn't even get a regulated return.

The margin trajectory confirmed the structural problem. T-Mobile's 2012 "Un-carrier" rebranding — using lower-cost AWS spectrum to undercut AT&T and Verizon on price — introduced competitive compression into a market that had enjoyed comfortable oligopoly pricing. ARPU fell industry-wide. The carriers responded the only way the infrastructure cycle permits: investing more capital to maintain competitive position, which increased the rate base without improving the return.

The carriers built the highway. Apple and Google built the cars. The highway collects tolls. The cars collect the GDP.

The 5G Bet

The 5G investment cycle has amplified every structural tension in wireless economics.

U.S. carriers have spent hundreds of billions on mid-band spectrum and 5G network buildout. The business case for this investment does not rest on consumer revenue — consumer ARPU is structurally capped by competition. It rests on enterprise services: private 5G networks for factories and warehouses, network slicing for latency-sensitive applications, IoT connectivity for industrial automation.

Whether carriers can successfully migrate from commodity consumer connectivity to differentiated enterprise services is the defining unanswered question of wireless economics. The structural pressure is identical to that facing every mature infrastructure: commodity infrastructure earns commodity returns. The only escape is migration up the value stack toward services with higher margins and higher switching costs. Cloud providers are attempting the same migration — from commodity IaaS to proprietary managed services. AT&T attempted the same migration 40 years ago — from commodity long-distance to proprietary equipment and services. The history of that attempt is not encouraging.

Compared to What

The cellular treadmill differs from the electrical utility treadmill in one critical respect. Electrical utilities face regulatory compression: the allowed return is set by a commission, it compresses gradually, and it has a floor — capital won't flow in at zero return, so the regulators maintain a return sufficient to attract investment. The compression is slow, predictable, and guaranteed to produce modest but survivable economics.

Wireless carriers face competitive compression: T-Mobile proved that a well-capitalized competitor could force industry-wide price cuts with no regulatory floor. The market sets the price. The market doesn't guarantee survival. Telecom faces what strategists have called the worst of both worlds — regulated like a utility in some respects (spectrum allocation, merger review, universal service obligations) but without the guaranteed ROE that at least gives electrical utilities predictable earnings.

Read the last row. That's the structural difference. The electrical utility built the grid and kept a regulated margin on it. The wireless carrier built the network and watched Apple, Google, and Netflix capture the economic value it created. Both are treadmills. One has a railing.

Sixty-Five Dollars

Sixty-five dollars was the monthly revenue per subscriber when the carrier was the product — when making a phone call was the service and the network was the value. Under forty-five dollars is the revenue now that the carrier is the pipe — the invisible layer between the consumer and the app, the thing you notice only when it doesn't work.

The $81 billion C-band auction was a bet that 5G enterprise services can reverse the dumb-pipe trajectory. The base rate from prior infrastructure cycles suggests that commodity infrastructure earns commodity returns regardless of the generation label — 3G, 4G, 5G, or whatever comes next. The carriers are spending as if differentiation is possible. The economics of every prior infrastructure era suggest it isn't — unless you own the application layer, not just the pipe. And the carriers, structurally, are not application companies. They are pipe companies arguing that this time, the pipe will capture the value. The sixty-five-dollar era is not coming back.

Part of the Infrastructure Economics series