• Question: how does random access memory work and why is there a shortage.

    Asked by austin to SamG, Penny T, johneaston on 20 Jan 2026.
    • Photo: John Easton

      John Easton answered on 20 Jan 2026:

      It’s actually quite simple, though the engineering is spectacular here. A memory chip consists of a number of memory ‘cells’. Each ‘cell’ is just a transistor and a capacitor. If there is charge in the capacitor then the memory bit is a “1”. If not, then it’s a “0”. The state of these capacitors is refreshed every few milliseconds to ensure that the charge remains in the correct state so that the data stored in the memory cells remains the same.

      That is the basic stuff. All the cool engineering comes when you look at modern enterprise-class memory. Reliability is built in (think of applying the same techniques as storage RAID to memory). Bits that fail can be ‘steered’ around meaning that a given memory chip can survive a number of hardware failures without needing to be replaced. As manufacturing techniques have improved and the size of memory cells get smaller, new ways to improve the stability of memory have been used by separating the memory cells vertically. So lots of cool things happen in something so fundamental to computing as a memory chip.

      Final thought for you. We’re starting to approach the limits of memory storage. So… how many atoms do you need to reliably hold 1 bit of information? Our researchers have proven that the answer is 11!

Comments

  • Photo: johneaston

    johneaston commented on :

    There’s a shortage of memory because so much is being consumed by the large datacentres being built to support the training of large AI models. As more and more memory goes to these, there is less available for other uses and consequently the price goes up.