Eniac
ENIAC, the Electronic Numerical Integrator and Computer, was a large general-purpose electronic digital computer built at the University of Pennsylvania's Moore School of Electrical Engineering for the US Army, designed chiefly by J. Presper Eckert and John Mauchly. It became operational in late 1945, was publicly unveiled in February 1946, and used about 17,468 vacuum tubes to calculate thousands of times faster than the electromechanical machines that came before it. It is remembered less as the first of anything than as the machine that made clear electronic speed was real, practical, and worth building at scale.
ENIAC was commissioned for a very concrete wartime problem: artillery firing tables. The Army's Ballistic Research Laboratory needed tables showing how to aim guns at different ranges and conditions, and producing each trajectory by hand or desk calculator took a person days. Eckert and Mauchly proposed an electronic machine in 1943, construction ran through 1944–45, and the first serious problem it ran, late in 1945, was a calculation for the Los Alamos weapons program rather than a ballistics table. After the public demonstration it was moved to Aberdeen Proving Ground in 1947 and stayed in service until it was switched off for the last time in 1955.
The scale of the machine is the easiest way to feel what vacuum-tube computing meant. ENIAC occupied about 1,800 square feet, weighed about 30 tons, drew roughly 150 kilowatts, and contained about 17,468 vacuum tubes plus thousands of relays, resistors, and capacitors. Its twenty accumulators each stored a ten-digit decimal number in ring counters, and it could perform about 5,000 additions per second — enormously faster than a human computer, and far faster than the relay machines of the time. It was decimal rather than binary, built from separate units for adding, multiplying, and storing, and the units were connected by cables carrying pulses between them.
ENIAC had no program in memory; to give it a new problem its operators set switches and re-plugged cables, and that work was done by a small team of women. Kay McNulty, Betty Jennings, Betty Snyder, Marlyn Wescoff, Fran Bilas, and Ruth Lichterman had been hired as human computers to calculate ballistics, and were selected to work out how to make the machine solve problems from diagrams and logic drawings, with no manuals or programming language to lean on. Their job was part mathematician, part electrician, and part debugger, and the techniques they developed for breaking a problem into steps and routing signals through the machine amount to some of the earliest programming. For decades their role was overlooked, and it is now widely recognized as a founding chapter of software.
The famous reliability worry about ENIAC was real but was managed rather than fatal. With thousands of tubes that all had to work simultaneously, the naive expectation was that the machine would be down constantly, and early operation did see frequent failures, with commonly cited accounts putting tube failures at about one every day or two. Eckert's design reduced the problem by running tubes well below their rated stress, minimizing power cycling since heating and cooling cycles killed many tubes, and testing components carefully. The lesson, that a computer of this size only works if its basic switch is dependable, is exactly why the industry later cared so intensely about replacing the tube.
ENIAC's rewiring problem helped motivate the stored-program idea, but the story is more nuanced than a clean replacement. In 1945 the EDVAC design report, associated with John von Neumann and the Moore School group, described a machine that held instructions in the same memory as data, and soon Manchester's Baby (1948) and Cambridge's EDSAC (1949) put the concept into operation. ENIAC itself was modified in 1948 to run in a form that read instructions from its function tables, which made programming much faster though somewhat slower to run. A related misconception is that ENIAC was the first computer; electronic precursors like the Atanasoff-Berry computer and Colossus came earlier, and a 1973 court decision found the ENIAC patent invalid, so the accurate claim is first large general-purpose electronic digital computer.
The legacy of ENIAC shows in everyday computing more than most people realize. Its size and tube count, with thousands of switches that all had to function, made the case for the transistor and the integrated circuit, and the people who worked on it carried ideas into UNIVAC and the early industry. A modern phone processor performs billions of operations per second per core on a chip with tens of billions of transistors, while consuming a few watts; ENIAC drew about 150 kilowatts, while the same arithmetic is now a rounding error on a chip.
Read ENIAC as the proof of concept that made electronic computing a practical bet: a room-sized, tube-filled, hand-rewired machine whose clumsiness taught the field what to fix, from the stored program to the reliable, tiny switch.