There is no single answer
People often ask who built the first computer, but historians do not agree on one machine, because the answer depends on which features are required. Must it be electronic? Must it be programmable? Must it keep its program in memory? Must it be general purpose? Each condition picks out a different machine. The most accurate answer is a sequence of milestones from several countries, some of them secret for years.
Babbage’s designs
Charles Babbage designed his Difference Engine in the 1820s, building on earlier calculating machines, to calculate and print mathematical tables, and a larger Analytical Engine in the 1830s that would be controlled by punched cards, with a separate store and arithmetic unit. The card idea came from the Jacquard loom, and Ada Lovelace’s notes on the engine are why she is often remembered as an early programmer. Neither was completed in his lifetime, partly because of the cost and the difficulty of making precision parts. Many historians regard the Analytical Engine as the first design for a general-purpose computer, but it was only a design. In 1991 the Science Museum in London completed a working Difference Engine from Babbage’s later plans, showing that the design worked.
Zuse’s Z3
In Berlin, Konrad Zuse built a series of machines in the late 1930s and early 1940s. His Z3, finished in 1941, used electromechanical relays and ran programs from punched tape. It is often described as the first working programmable, automatic digital computer. It lacked the conditional jump that makes a machine fully general, though later analysis suggested it could be used for general computation with a roundabout method. The original was destroyed in a 1943 bombing raid, and a reconstruction was built later.
The Atanasoff-Berry Computer
John Atanasoff and Clifford Berry at Iowa State College built a prototype in 1939 and a larger machine by about 1942. It used vacuum tubes for arithmetic and was designed to solve systems of linear equations. It was electronic and digital but not programmable in a general sense. In a 1973 US court decision about the ENIAC patent, the judge found that ENIAC’s designers had derived some ideas from Atanasoff, which cast doubt on the patent and raised the profile of his work. Historians still argue about how much influence it actually had.
Colossus
At Bletchley Park in Britain, engineer Tommy Flowers led the building of Colossus, with the first machine working in late 1943 and an improved version in 1944. It used thousands of vacuum tubes to help read German teleprinter ciphers. It was electronic and could be configured with switches and plugs for different tasks, but it was built for a specific purpose and not for general calculation. Its existence stayed secret for decades, which is why it was long left out of popular histories.
Harvard Mark I and ENIAC
The Harvard Mark I, completed in 1944 under Howard Aiken with IBM’s help, was a large electromechanical calculator that read instructions from paper tape. ENIAC, built at the University of Pennsylvania and publicly announced in 1946, was a general-purpose electronic machine, far faster than mechanical ones. It was programmed by setting switches and rewiring, which could take days. For years ENIAC was described as the first electronic computer in many accounts, but the evidence about Colossus and the Atanasoff-Berry machine has qualified that claim, and the term is better read as the first large general-purpose electronic digital computer.
The Manchester Baby
The idea of a stored program, with instructions kept in the same memory as data, was discussed in a 1945 report by John von Neumann, and builds on work by many people. The first machine to run a program from electronic memory was the Small-Scale Experimental Machine, nicknamed the Baby, at the University of Manchester. Frederic Williams, Tom Kilburn and Geoff Tootill built it, and it ran its first program on June 21, 1948, with memory stored as spots of charge on cathode ray tubes. It was small and meant to test the memory technology. A fuller Manchester computer followed, and EDSAC at Cambridge entered regular service in 1949.
Why the dispute continues
Different definitions give different winners. If a first computer must be mechanical and programmable in design, Babbage is the answer. For working programmable digital calculation, Zuse’s Z3 is a strong candidate. For electronic digital calculation, the Atanasoff-Berry machine and Colossus both qualify with caveats. For general-purpose electronic machines, ENIAC. For the stored-program machine that resembles today’s computers, the Baby and EDSAC. Another complication is that wartime secrecy hid some projects, and national pride has shaped how the story is told.
A practical view
A balanced summary is that the computer was developed in stages, with ideas moving between people, and that no single “eureka” moment exists. The stored-program architecture is the strongest link to modern machines, which is why the Baby carries so much historical weight despite its small scale.