The Bell Wars · Ethan Ding

Each ribbon is network (in share points of the market) passing from the absorbed to the absorber. Red ribbons = destroyed — a court's decree, an $11B fraud, a forced spinoff. Gold ribbons = merger or acquisition. The chart contains the two most extraordinary ribbons in this whole series: the eight-way shatter of the Bell System in 1984 — and the thin gold ribbon in 2005 where AT&T flows into the child that bought it. Trace also the tragic loop: AT&T spins off its wireless arm in 2001, and buys it back — via its siblings — three years later, at a $30B markup. Click any ribbon or carrier node to pin the deal or decree behind it.

Why the fourth-place carrier won. Three structural advantages, compounding for a decade. (1) Its enemy paid for its revival: when the DOJ blocked AT&T's $39B purchase of T-Mobile in 2011, the contract's breakup package — $3B in cash, plus billions in spectrum and a national roaming agreement — flowed to the target. The largest breakup fee in corporate history capitalized the resurrection of the exact rival that would spend the next decade dismantling AT&T's wireless margins. Hideyoshi arming the Tokugawa has nothing on this. (2) Nothing to lose, so it burned the rulebook: John Legere's "Un-carrier" (2013) abolished, one by one, every profitable heresy the duopoly depended on — contracts, overage fees, international roaming charges, upgrade cycles. Fourth place could attack the industry's margin pool because it barely participated in it; AT&T and Verizon had to choose, every time, between matching and bleeding. (3) A patient foreign sovereign: Deutsche Telekom — majority state-influenced, thinking in decades — held through years when every analyst demanded a sale. The same patient-capital advantage Japan's banks gave the DRAM makers and Riyadh gave Aramco, applied to Bellevue, Washington.

The conquest. In 2020 T-Mobile captured Sprint — after Sprint's own SoftBank era failed to fix it — and with it the 2.5 GHz mid-band spectrum trove: the Kantō plain of 5G, the one resource that let T-Mobile build the best national 5G network years before its rivals could assemble the equivalent. The carrier that was nearly sold for parts twice now trades places with Verizon for the wireless crown, and the duopoly it mocked is a triopoly it anchors.

Why the aura is already being tested. Un-carriers age into carriers: price increases on "never increase" plans, the John Legere theater retired, and the margin pool it once raided now substantially its own. The pattern from every chart in this series — Southwest, above all — suggests what comes next; the only question is which fourth place is being funded, right now, by the breakup fees of the future. (The satellite band in gray has a candidate.)

The strangest fact about the largest corporate breakup in history is that AT&T picked which half to keep — and picked wrong on every axis. Facing Judge Greene, the company could shape the settlement, and it chose to keep the crown jewels as it understood them: long distance (the profit engine), Western Electric (the world's greatest factory), Bell Labs (the transistor, information theory, Unix — the greatest research institution ever assembled), and its computing ambitions. It surrendered what it considered the boring part: seven regional companies of regulated local copper. The mother kept Mikawa and handed the children the Kantō.

Every assumption inverted within a decade: (1) long distance, the treasure, became a commodity knife-fight — MCI's "Friends & Family" and Sprint's pin-drop fiber drove margins toward zero; (2) the boring local loop turned out to be the bottleneck asset — the physical connection to every home became DSL, then broadband; (3) and the decree's quietest clause was the largest transfer of the century: the cellular licenses went with the regions — to the children. The mother of the telephone was born into the wireless age owning no wireless at all, bought McCaw for $12.6B in 1994 to fix it, spun that arm off at the very top in 2001, and watched its siblings' venture buy it for $41B three years later. Meanwhile Western Electric and the Labs — the jewels — were themselves spun off (Lucent, 1996), rode the bubble, crashed 99%, and ended as a subsidiary of Nokia. Every asset AT&T chose to keep was eventually worth less than any asset it gave away.

The regranting completed the inversion. In 2005, Southwestern Bell — at divestiture among the least glamorous of the seven children — bought its mother for $16 billion (roughly what AT&T had lost on cable television alone) and, in the era's perfect gesture, took her name. BellSouth followed for $86B; six of 1984's eight fragments now live inside AT&T Inc. and Verizon. And then the reassembled empire proved that institutions repeat their founding mistake: convinced once again that the future lay in content rather than the network, the new AT&T spent $67B on DirecTV and $108B on Time Warner — and divested both within seven years at a combined loss north of $150B. The exile taught the children everything; it taught the name nothing.

On January 1, 1984, the Bell System's local networks were divided among seven regional holding companies of deliberately equal weight — the bars show access lines at divestiture. Wall Street called them the boring half of the breakup. Within 22 years, six of the seven had consolidated into two empires — and one of those two had purchased the mother herself. The seventh wandered into fraud, three renames, and ruin. Hover a bar for each regent's fate.

The era's major powers, in order of when they folded. "Folded" takes three forms: destroyed (broken by decree, fraud, or bankruptcy), absorbed (merged into a survivor), or still connected in 2025. Note the signature of this industry: like oil, the great falls were inflicted by the state (1913, 1956, 1984, the blocked mergers) — but unlike any other chart, the shattered empire substantially reassembled itself, and the map of 2025 is more concentrated in wireless than regulators of 1984 would ever have permitted in wireline.

How the share estimates were made (Fermi method): Anchors are FCC statistics, census data and company reports. Documented checkpoints keep the bands honest: the independents held roughly half of America's telephones by 1907 (though less of its revenue — Bell held the cities and the toll lines); the Bell System carried ~80–85% of the nation's telephones through the regulated century and employed ~1 million people at its peak — the largest private employer on Earth; AT&T took ~90% of long distance at divestiture and barely half within a decade; WorldCom's fraud was $11B and its bankruptcy the largest in history (2002, a record it held for six years); SBC paid $16B for AT&T (2005) and $86B for BellSouth; T-Mobile's Sprint merger (2020) created today's wireless triopoly, in which the three carriers split ~97% of postpaid subscribers. Bands include each carrier's wireless, wireline and broadband service revenue; media and equipment segments are excluded where separable. Error bars are ±2 points, wider before 1930.

Two things that look wrong but aren't. (1) The Bell band understates the monopoly — as always with the mothers. Like Standard, Fairchild and the Pennsylvania before it, the Bell System's true power sat outside the revenue measure: it was a private government — its own manufacturer (Western Electric), its own research state (Bell Labs), rates set like taxes, and a workforce the size of an army. The 1956 consent decree that forced it to license every patent royalty-free is arguably the most consequential antitrust remedy ever issued: those transistor licenses seeded the entire semiconductor chart. (2) The gray band is three different rebellions wearing one color. Before 1913 it is the 6,000 independents of the first open era; in the 1990s it is the CLECs the Telecom Act briefly conjured; today it is Dish's failed fourth network, US Cellular's remnant, the rural carriers — and Starlink, which is the 1894 patent expiry happening again, from orbit. Every closed system in this series eventually meets its independents' war; the gray band is where the next chart begins.