Major upgrades of the U.S. telephone network have taken decades. The first U.S. commercial installation of a automatic (mechanical) telephone switch occurred in 1892. In the Bell System in 1920, nearly three decades later, only 2% of telephones were automatically switched. The share of telephones in the U.S. that were automatically switched didn’t rise to 75% until 1950. The category “electronic” switch was added to FCC statistics in 1968. Two decades later, the share of telephones that were electronically switched first rose above 75%. 
Implementing these new telephone network technologies required coordinated upgrades in central office equipment and end-user equipment. The development of more powerful, more general purpose end-user equipment allows central (cloud) service upgrades to be more independent of end-user equipment upgrades. That independence contributes to more rapid network innovation.
U.S. telephone companies owned most of the telephones made obsolete by upgrades to mechanical and electronic switching. Technical change that makes obsolete user-owned equipment and requires users to buy more equipment is more likely to be commercially favored. User frustration and anger puts some constraint on forced upgrades. Nonetheless, user ownership of end-user equipment provides greater network operator incentives for network service upgrades.
Upgrades of telephone network technology occurred faster in larger exchanges. In 1936, non-common-battery exchanges, which were very technologically backward, served on average 180 telephones per exchange. Mechanically switched central offices served on average about 5900 telephones, more than twice as many as the average for manually switched central offices. A higher average number of telephones per exchange correlated regionally with a higher share of mechanically switched telephones. Larger exchanges’ lead in upgrades continued with the upgrade from mechanical to electronic switches. In 1980, electronic central offices served on average 31% more telephones than did mechanically switched central offices.
Fixed costs of upgrades and greater service growth favored faster upgrades in larger exchanges. A significantly share of exchange upgrade costs were likely not highly correlated with the number of telephones served. Hence an exchange with more telephones had a lower upgrade cost per telephone (and telephones were essentially the revenue unit). In addition, because population was shifting from farms to towns and cities, larger exchanges were in places that were growing more rapidly in population. Potential for future subscriber growth favored larger exchanges.
Company size was not strongly associated with the the share of mechanically switched telephones in 1939. All companies serving more than 25,000 telephones had some mechanically switched telephones. Many smaller companies had no mechanically switched telephones. But small companies pioneered mechanical switching. Some relatively small companies had relatively high shares of mechanically switched lines. Among companies serving more than 1000 telephones and having some mechanically switched telephones, the share of mechanically switched lines was not significantly correlated with company size.
Relative to company size, industry structure deserves more attention in thinking about how to foster innovation. General-purpose devices, end-user device ownership, and common technology knowledge readily available to small organizations all contribute importantly to faster network innovation.
* * * * *
 The first mechanical telephone switch in the U.S. was installed in La Porte, Indiana. It used a Strowger switch. About this time telephone systems began implementing common-battery systems. These systems powered end-user telephones with a battery at the central office. In 1936, 5% of U.S. telephones were not powered from a common source in the central office.
 Lipartito, Kenneth James, “When Women Were Switches: Technology, Work and Gender in the Telephone Industry, 1890-1920,” American Historical Review, October, 1994, p. 1095, notes that in 1915 about 400,000 independent lines were automatically (mechanically) switched. Independents served roughly one-third of total U.S. telephones. In 1915, the Bell System had no automatically (mechanically) switched telephones.
 Other categories of switches separate from “electronic” were “manual,” “step-by-step dial,” “cross-bar dial,” “panel dial,” and “other (rotary and relay dial)”. “Electronic” seems to mean a switch using transistor (semi-conductor) circuits.
 Here’s the individual telephone company data for 1939. Among the five telephone companies with greater than 90% of telephones mechanically switched, three served less than 10,000 telephones.