From pebbles on a counting board to billions of switches on a chip: 4,500 years of teaching machines to follow instructions.
People have always used tools to help them count, from pebbles and beads to boxes of gears. In the 1800s Babbage and Lovelace imagined a machine that could follow any list of instructions, and in the 1930s Turing and Shannon showed how logic and switches could do it. Then came room-sized electronic machines, the transistor, the chip and the personal computer. India built its own computers from 1960 and its own supercomputers from 1991, and today computers sit in every pocket, doing many things at once.
Beads, pebbles and gears help people add and carry, but a person still has to do the thinking.
2500 BCE
c. 2700–2300 BCE
Counting with pebbles and beads
Unknown merchants and scribesMesopotamia; later Greece and China
People in ancient Mesopotamia are thought to have counted on boards with lines and pebbles. The oldest counting board that survives, the Salamis tablet from Greece, dates to about 300 BCE. Written records of the Chinese abacus go back to about the 2nd century BCE.
Why it mattered. Each bead or pebble stands for a digit in a place, the same idea a computer uses for numbers today.
A 19-year-old Blaise Pascal built a calculator to help his father, a tax official, with long sums. Turning dials moved gears, and a clever carry mechanism passed a 1 to the next wheel. About 50 were made over the next ten years.
Why it mattered. It showed that a machine, not a person, could do the carrying in arithmetic.
Punched cards, Babbage's engines, Lovelace's notes, Boole's logic and Turing's machine turn the idea of a list of instructions into a science.
1804
A loom that follows punched cards
Joseph Marie JacquardLyon
Jacquard's loom wove patterns by reading a chain of cards with holes punched in them. A hole meant lift this thread, no hole meant leave it. Change the cards and the same loom weaves a different pattern.
Why it mattered. Holes and no holes are a kind of 1s and 0s, and the chain of cards was a stored list of instructions.
Charles Babbage designed a Difference Engine to print mathematical tables without human mistakes. By 1837 he had designed something far bigger: the Analytical Engine, with a 'store' for numbers and a 'mill' to work on them, fed by Jacquard-style punched cards. Neither was finished in his lifetime.
Why it mattered. The store and the mill are the memory and the processor of every computer since.
Ada Lovelace translated an article about the Analytical Engine and added notes three times longer than the article. Her Note G set out, step by step, how the engine could calculate Bernoulli numbers. It is widely called the first published computer program.
Why it mattered. She also saw that such a machine could work with symbols and even music, not just numbers.
George Boole, a self-taught English mathematician working in Ireland, published The Laws of Thought. He showed that true and false could be handled like numbers, with rules for AND, OR and NOT.
Why it mattered. Boole's algebra of true and false is exactly what a computer's logic gates do.
In his paper 'On Computable Numbers', 24-year-old Alan Turing imagined a simple machine that reads and writes symbols on a long tape, following a table of rules. He showed that one universal machine could imitate any other, just by being given its rules.
Why it mattered. It is the idea behind every computer: one machine, many programs.
In his master's thesis, Claude Shannon showed that circuits of on/off switches follow Boole's algebra. A switch that is on is 1, a switch that is off is 0, and wiring them together can add numbers.
Why it mattered. It joined logic to electricity, the step that turned Boole's ideas into hardware.
Grace Hopper wrote the A-0 system, an early compiler: a program that turns instructions written by people into machine code. Many experts told her computers could only do arithmetic. Her later work led to the business language COBOL.
Why it mattered. Compilers let people write programs in words instead of raw numbers.
Room-sized machines full of vacuum tubes break codes, compute tables and learn to keep their programs in memory.
1944
1943–1944
Colossus breaks codes
Tommy Flowers and the Bletchley Park codebreakersDollis Hill and Bletchley Park
During the Second World War, engineer Tommy Flowers built Colossus to help crack the German Lorenz cipher. It used about 1,600 vacuum tubes in its first version and was working by early 1944. It was kept secret for decades.
Why it mattered. It is often called the first programmable, electronic, digital computer, though it was built for one kind of job.
J. Presper Eckert and John MauchlyUniversity of Pennsylvania, Philadelphia
ENIAC filled a large room: about 18,000 vacuum tubes, 30 tons and 150 kilowatts of power. It could do about 5,000 additions a second, far faster than anything before it. It was finished in 1945 and shown to the public in February 1946.
Why it mattered. It proved that a big, general-purpose electronic computer could really work.
John von Neumann, with the ENIAC teamMoore School, Philadelphia
John von Neumann wrote the 'First Draft of a Report on the EDVAC'. It described a computer that keeps its program in the same memory as its data, so a new program can be loaded as easily as new numbers. Eckert and Mauchly had worked on these ideas too, and the credit is still argued about.
Why it mattered. Almost every computer since has kept its program in memory this way.
Kay McNulty, Betty Jennings, Betty Snyder, Marlyn Wescoff, Fran Bilas, Ruth LichtermanUniversity of Pennsylvania, Philadelphia
ENIAC had no keyboard and no programming language. Six women worked out how to program it by setting switches and plugging cables, often from wiring diagrams alone. At the 1946 unveiling they were barely mentioned.
Why it mattered. They were among the first programmers, and their story was left out of most histories for decades.
Frederic Williams, Tom Kilburn and Geoff TootillUniversity of Manchester
A small test machine nicknamed the Baby ran its first program on 21 June 1948. The program, which found the largest factor of a number, sat in the machine's own electronic memory, a special cathode ray tube.
Why it mattered. It was the first computer to run a program stored in its own electronic memory.
The transistor and the chip shrink the computer from a room to a desk, and Moore's law keeps it shrinking.
1947
16 December 1947
The transistor
John Bardeen, Walter Brattain and William ShockleyBell Labs, Murray Hill, New Jersey
At Bell Labs, Bardeen and Brattain made a tiny switch and amplifier from a piece of germanium. Shockley soon designed a better version. The three shared the 1956 Nobel Prize in Physics.
Why it mattered. Transistors replaced hot, fragile vacuum tubes and are the switches inside every chip today.
Jack Kilby (Texas Instruments) and Robert Noyce (Fairchild)Dallas, Texas and Mountain View, California
In 1958 Jack Kilby built several parts of a circuit on one piece of germanium. In 1959 Robert Noyce patented a way to make whole circuits on flat silicon, with metal wires printed on top. Kilby won the 2000 Nobel Prize in Physics, and both are credited as inventors.
Why it mattered. Putting many transistors on one chip made computers smaller, cheaper and faster every year.
In a magazine article, Gordon Moore noticed that the number of parts on a chip was doubling every year, and guessed it would continue. In 1975 he changed his estimate to doubling about every two years.
Why it mattered. The prediction became a target the whole chip industry raced to hit for 50 years.
Federico Faggin, Ted Hoff, Stanley Mazor and Masatoshi ShimaIntel, Santa Clara, California
The Intel 4004 put a whole processor on one chip, with about 2,300 transistors running at up to 740 kHz. It was first designed for a Japanese company's calculator. It is widely called the first commercial microprocessor.
Why it mattered. Every phone and laptop today has descendants of this chip at its heart.
The Apple II came ready to use, in a plastic case with a keyboard, and could show colour on a TV. It was one of three popular home computers of 1977, along with the Commodore PET and TRS-80.
Why it mattered. Computers moved from company basements into homes and schools.
IBM launched its Personal Computer, built mostly from parts other companies made, including an Intel processor and Microsoft's DOS. Other firms soon built 'clones' that ran the same software.
Why it mattered. Its open design became the standard for most desktop computers for decades.
Each step up is ten times more. In about 50 years, chips went from thousands of transistors to hundreds of billions.
1971 Intel 4004: about 2,300
1978 Intel 8086: 29,000
1985 Intel 80386: 275,000
1989 Intel 80486: about 1.2 million
1993 Intel Pentium: 3.1 million
2000 Intel Pentium 4: 42 million
2006 Intel Core 2 Duo: 291 million
2011 Intel Core i7 (Sandy Bridge): 1.16 billion
2017 Apple A11 phone chip: 4.3 billion
2020 Apple M1: 16 billion
2023 Apple M2 Ultra: 134 billion
2024 NVIDIA Blackwell B200: 208 billion, on two dies in one package
1955 – today
India builds its own
From TIFRAC and ISIJU to ECIL's factory computers, PARAM and today's AI supercomputers.
1960
February 1960; named by Nehru c. 1962
TIFRAC, India's first computer
Rangaswamy Narasimhan and team, TIFRBombay (now Mumbai)
A team at the Tata Institute of Fundamental Research built India's first home-made digital computer, based on the IAS machine design from Princeton. It used about 2,700 vacuum tubes and an early ferrite core memory. It was commissioned in 1960, and Prime Minister Nehru gave it its name; sources differ on whether that was in 1960 or 1962.
Why it mattered. It showed that India could design and build its own computers.
Samarendra Kumar Mitra and Jnan Saran ChatterjeeIndian Statistical Institute and Jadavpur University, Calcutta (now Kolkata)
Engineers from the Indian Statistical Institute and Jadavpur University built ISIJU-1, a computer made with transistors instead of vacuum tubes. It was working by about 1964 and was formally commissioned on 2 April 1966.
Why it mattered. It was India's first solid-state digital computer.
S. Srikantan and team, BARC and ECILTrombay (Bombay) and Hyderabad
A prototype 12-bit computer, the TDC-12, was completed in 1969 and made by the new public company ECIL. About 20 had been supplied by 1971, for jobs like watching for earthquakes. ECIL went on to make the 16-bit TDC-316 and, in 1979, the 32-bit System 332.
Why it mattered. India began making computers in a factory, not just in a lab.
When India could not buy a big American supercomputer, the government set up C-DAC in 1988 to build one. Its answer, PARAM 8000, linked many small processors to work in parallel. It was rated at about 1 gigaflop, a billion calculations a second, at its peak.
Why it mattered. It proved that many small processors working together could match a costly supercomputer.
C-DAC and the National Supercomputing MissionPune and across India
PARAM Siddhi-AI ranked 63rd in the world in November 2020, at about 4.6 petaflops. In 2023 the AIRAWAT system at C-DAC ranked 75th. By August 2025 the National Supercomputing Mission, begun in 2015, had installed 37 supercomputers with about 40 petaflops in total.
Why it mattered. India now builds and runs supercomputers for weather, medicine and AI research.
The web, GPUs, many-core chips and smartphones put billions of computers in people's hands, all doing many things at once.
1991
1989–1991
The World Wide Web
Tim Berners-LeeCERN, near Geneva
Tim Berners-Lee proposed a way to link documents across computers in 1989. By 1991 the web was open to people outside CERN, and in 1993 CERN put the software into the public domain.
Why it mattered. It turned computers from calculating machines into windows onto the world.
NVIDIA announced the GeForce 256 and marketed it as the world's first GPU, a chip that did the maths for 3D graphics itself instead of leaving it to the main processor. It had about 17 million transistors.
Why it mattered. GPUs do thousands of small sums at once, which later made them perfect for AI.
IBM's POWER4 put two processor cores on a single chip, the first mainstream processor to do so. AMD and Intel brought dual-core chips to ordinary PCs in 2005.
Why it mattered. When clock speeds stopped rising, computers got faster by adding more cores instead.
Apple, presented by Steve JobsSan Francisco, California
Apple showed the iPhone, a phone with a touchscreen, a web browser and a full operating system. The first Android phone followed in 2008. Smartphones soon became the most common computers on Earth.
Why it mattered. For billions of people, their first computer is a phone.
Alex Krizhevsky, Ilya Sutskever and Geoffrey HintonUniversity of Toronto
In 2006 and 2007 NVIDIA released CUDA, which let programmers use graphics chips for any kind of maths. In 2012 a neural network called AlexNet, trained on two gaming GPUs, won an image recognition contest by a wide margin.
Why it mattered. It started the modern AI boom, which runs on huge numbers of chips working in parallel.
ENIAC used about 150 kilowatts, as much as about 75 electric kettles boiling at once, to do 5,000 additions a second. A phone chip today does billions of operations a second on a few watts.
Ada Lovelace wrote about a computer program about a hundred years before anyone built a computer that could run it.
The Intel 4004 had about 2,300 transistors. NVIDIA's Blackwell B200 of 2024 has 208 billion, about 90 million times more.
The word 'computer' used to mean a person, often a woman, whose job was to do calculations by hand.
India's first supercomputer, PARAM 8000, was built because India could not import one from the USA.
The people
Who figured it out
CB
Charles Babbage
1791 – 1871 · Mathematician and inventor · England
Designed the Analytical Engine, a mechanical computer with a store and a mill.
AL
Ada Lovelace
1815 – 1852 · Mathematician and writer · England
Wrote the first published program and saw that computers could handle more than numbers.
AT
Alan Turing
1912 – 1954 · Mathematician · England
Imagined the universal machine and helped break codes in the Second World War.
KZ
Konrad Zuse
1910 – 1995 · Engineer · Germany
Built the Z3 in Berlin in 1941, a working programmable computer made with relays.
KM
Kay McNulty
1921 – 2006 · ENIAC programmer · Ireland and USA
One of the six women who first programmed ENIAC, by plugging cables and setting switches.
JJ
Jean Jennings Bartik
1924 – 2011 · ENIAC programmer · USA
Betty Jennings in 1946; she later helped turn ENIAC into a stored-program machine.
JN
John von Neumann
1903 – 1957 · Mathematician · Hungary and USA
His 1945 report described the stored-program design most computers still use.
GH
Grace Hopper
1906 – 1992 · Computer scientist and US Navy officer · USA
Built an early compiler and pushed for programs written in words people can read.
JK
Jack Kilby
1923 – 2005 · Engineer · USA
Made the first integrated circuit at Texas Instruments in 1958.
RN
Robert Noyce
1927 – 1990 · Engineer and co-founder of Intel · USA
Invented the flat silicon integrated circuit that chips are still made on.
GM
Gordon Moore
1929 – 2023 · Chemist and co-founder of Intel · USA
Predicted that chips would keep doubling their transistors.
RN
Rangaswamy Narasimhan
1926 – 2007 · Computer scientist · India
Led the TIFR team that built TIFRAC, India's first home-made computer.
SK
Samarendra Kumar Mitra
1916 – 1998 · Scientist and mathematician · India
Built an analog computer in 1953–54 and led the ISIJU-1 transistor computer.
VB
Vijay Bhatkar
born 1946 · Computer scientist · India
Founding director of C-DAC, which built the PARAM supercomputers.