Before electronic machines
The word “computer” first described a person who carried out calculations by hand, often in teams working on tables for astronomy, navigation or artillery. Mechanical aids existed long before electronics. Gear-driven mechanical calculators could add and multiply from the 1600s onward, and by the 1800s they were sold to businesses.
The big conceptual step came from Charles Babbage, an English mathematician who designed a general-purpose Analytical Engine in the 1830s. It was never finished in his lifetime, but the design had a processing unit, a store for numbers and instructions fed in on cards, an idea borrowed from the Jacquard loom. Ada Lovelace wrote notes on the engine that included a method for calculating a sequence of numbers, and she is often called the first programmer, though historians debate how much of the program was her own work.
The first electronic computers
Electronic machines arrived in the 1940s, mostly driven by wartime needs. Several projects ran in parallel in Britain, Germany and the United States, and which one counts as “first” depends on the definition used. The machine most people name is ENIAC, built at the University of Pennsylvania by J. Presper Eckert and John Mauchly and unveiled publicly in 1946. It used thousands of vacuum tubes, filled a large room and was designed to calculate artillery firing tables. It was programmed by rewiring cables and setting switches, a job done largely by a team of women who became its first expert programmers.
The next shift was the stored-program idea: keeping instructions in the same memory as data, so that changing a program meant loading new contents instead of rebuilding wiring. The Manchester “Baby” ran a small stored program in 1948, and EDSAC in Cambridge began regular service in 1949. UNIVAC I, delivered in 1951, was among the first computers sold commercially in the United States. Credit for the stored-program concept is shared among several people, and a 1945 report by John von Neumann helped spread it.
Transistors and the mainframe era
Vacuum tubes were bulky, hot and prone to failure. The transistor, demonstrated at Bell Labs in 1947 by John Bardeen, Walter Brattain and William Shockley, did the same switching job in a much smaller, cooler and more reliable form. By the late 1950s, transistorized computers were replacing tube machines.
Then came the integrated circuits that packed many transistors onto one piece of silicon, developed independently in the late 1950s. Chips made computers cheaper and faster, and in 1964 IBM announced its System/360, a family of compatible machines that let customers upgrade without rewriting all their software. Through the 1960s computers were mostly large, expensive systems in companies, universities and government offices, shared by many users through batch jobs or time-sharing terminals.
From minicomputers to personal computers
Smaller minicomputers, such as the PDP series from Digital Equipment Corporation, brought computing to laboratories and small organizations. In 1971, Intel’s 4004 put a processor on a single chip, and the microprocessor made a personal machine thinkable. Hobbyist kits like the Altair 8800 appeared in 1975. In 1977 several ready-to-use machines, including the Apple II, reached buyers who were not engineers. The IBM Personal Computer of 1981 set a widely copied standard for business use, and the Macintosh of 1984 brought a graphical interface and mouse to a mass audience.
Networks and the web
Computers became far more useful when they were connected. Research networks in the late 1960s grew into the internet, a story told in more detail in the history of the internet. In the early 1990s Tim Berners-Lee’s World Wide Web and graphical browsers made the network easy for ordinary people to use. Email, online shopping and search changed how businesses and households used their machines, and laptops let people take them along.
Phones, cloud and data
In the 2000s, processors became efficient enough to fit into phones. Modern smartphones are touch-driven computers with cameras, sensors and constant network access, and they now outnumber desktop machines in everyday use. Meanwhile, large data centers began offering computing as a service, which is the basis of cloud computing. Companies and researchers also gathered far more information than before, which pushed demand for faster storage and processing.
Artificial intelligence
Researchers have worked on artificial intelligence since at least the 1950s. The term was used for a 1956 workshop at Dartmouth College, and early programs played games and proved simple theorems. Progress came in waves, with periods of enthusiasm followed by disappointment when results fell short of expectations. In the 2010s, a combination of large datasets, faster parallel processors and improved neural network methods produced major gains in image recognition, speech and translation. More recently, large language models have made systems that generate text and images widely available. Experts disagree about how capable these systems are, how they should be used and what comes next, so it is wise to treat predictions with caution.
Software and ways of using computers
Hardware is only half the story. Early machines were programmed in numeric machine code, which was slow and error-prone. In the 1950s, assemblers and then higher-level languages such as FORTRAN, introduced by IBM in 1957, and COBOL, designed around 1959, let people write instructions closer to ordinary mathematics and business terms. Grace Hopper was a leading advocate of the idea that programs could be written in readable words and translated by a compiler.
Ways of using machines changed as well. Batch processing, where a job was submitted and results collected later, gave way in the 1960s to time-sharing, in which many people used one machine at once from separate terminals. Graphical interfaces, developed in research labs in the 1960s and 1970s and sold widely from the 1980s, replaced typed commands for most users. Each of these shifts widened the circle of people who could make use of a computer without being specialists.
What stays constant
Across eight decades, the pattern is similar: a new component, such as the transistor, the chip or the parallel processor, makes computing cheaper and smaller, which brings in new users and new uses. Some researchers are now exploring quantum computing, which aims to solve certain problems in a different way, but it remains largely experimental and its practical reach is still uncertain.