The densest hard drive on Earth is already inside you.
Every one of your cells carries a full instruction manual written in just four chemical letters. Scientists can now use those same letters to save photos, films and books. This page shows how it works, how small it really is, what it costs today, and which claims are still hype.
The word DNA written three ways: as text, as binary, and as DNA letters.
From a whole person down to a single bit
Press the plus button to dive in, or drag the slider. Each step takes you deeper, from your body to the cell, to the nucleus, to the chromosome, to the twisted ladder of DNA, and finally to the data hiding in its rungs.
How 1010 becomes ACGT
Computers store everything as ones and zeros. DNA stores everything as four letters: A, C, G and T. Four options means each letter can stand for exactly two bits. That is the whole trick.
This is the simplest possible code. Real systems use smarter codes that avoid long repeats such as AAAAAA and add error correction, because DNA machines make small mistakes.
Works in any language, including Arabic and Hindi, and with emoji. Up to 40 characters.
How much DNA are you carrying right now?
More than you would guess, and almost all of it is the same file copied again and again.
Each cell with a nucleus holds two full sets of your genome, one from each parent.
Folded so tightly that it fits in a nucleus about 0.006 mm wide.
One set of the human genome is about 3.2 billion letters. At 2 bits each, that is roughly one CD of data.
You have 30 to 37 trillion cells, but red blood cells, the most common kind, carry no DNA at all.
Nature chose backups over capacity
Your body uses this incredible storage medium to keep trillions of copies of one small file, so every cell has its own complete manual. If those same grams were filled with different data instead, they could in theory hold more than 4 exabytes, which is over 4 million one terabyte drives.
What could a speck of DNA hold?
In 2017, researchers at Columbia University and the New York Genome Center showed a coding method that reaches a theoretical 215 petabytes per gram. Drag the slider to change the amount of DNA.
Theoretical maximum. Real systems store less because they add extra copies and error correction.
How long does each storage type last?
Typical useful life when stored well. The scale is logarithmic, so each step to the right is ten times longer.
Scientists have read DNA from mammoth teeth more than a million years old. Kept cold, dry and dark, DNA easily outlives every device we have ever built. It also never becomes an obsolete format, because people will always need to read DNA for medicine.
All the data in the world, in one small car
Humanity creates roughly 175 zettabytes of data a year. At 215 petabytes per gram, that would fit in about 800 kilograms of DNA. Today the same data needs warehouse sized data centres that use huge amounts of electricity and must be rebuilt every few years.
Writing and reading a DNA file
Five steps, in order. Tap a step to see what happens, or run them all.
The honest part: cost and speed
The science works. The economics do not, yet. Turning 1010 into ACGT is free software. The money goes into physically building the DNA, one letter at a time.
to write and read just 2 megabytes in the famous 2017 experiment. About 7,000 dollars to build the DNA and 2,000 to read it.
estimated cost to write one gigabyte with that same method. Magnetic tape writes a whole terabyte for around 10 dollars.
taken by the first fully automatic DNA storage machine, built by Microsoft and the University of Washington in 2019, to write and read the word hello.
is roughly how much more DNA costs per megabyte than an SSD today. Costs must fall by five to six powers of ten to compete with tape.
Four ways the cost can come down
Enzymes instead of harsh chemistry
Today DNA is built with a decades old chemical process that uses toxic solvents. New methods use enzymes, the same kind of tiny machines your cells use, and promise to be 10 to 100 times cheaper and cleaner.
Millions of strands at once on a chip
Instead of one strand at a time, silicon chips can grow millions of strands side by side. Storage does not need perfect DNA, so the process can be fast and sloppy, and software fixes the errors afterwards.
Smarter codes that use fewer letters
Better maths packs more bits into each letter and needs fewer spare copies. Fewer letters to build means a smaller bill for the same file.
Ready made blocks, like movable type
Some teams skip building every letter from scratch. They keep a library of short ready made DNA pieces and join them together, the way a printing press arranges existing letters instead of hand writing each page.
True or false?
Videos about DNA storage often skip the details. Tap each card to check the claim.
Seventy years from discovery to data
Check what you learned
Six quick questions. You get the explanation after every answer.
For classrooms and curious minds
Try this with paper and four coloured pens
- Pick a short word. Look up each letter's 8 bit binary code, or use the tool above.
- Split every 8 bits into four pairs.
- Swap each pair for a letter: 00 is A, 01 is C, 10 is G, 11 is T. Use a different colour for each.
- Swap papers with a friend and decode each other's word.
- Now change one letter on purpose. What happens to the message? This is why real systems add error correction.