There is something almost poetic about how we store the sum of human knowledge. Before the cloud, before the smartphone, before even the personal computer as we know it, engineers faced a deceptively simple question: how do you teach a machine to remember? The answer, it turns out, began with magnets the size of peppercorns and has evolved into technologies that heat microscopic spots to temperatures rivaling molten steel. This is the story of magnetic storage, and it reveals far more about human ingenuity than any specification sheet ever could.
When Computers Learned to Remember: The Dawn of Ferrite Core Memory
I have spent years studying the early days of computing, and what continually strikes me is how revolutionary the simplest solutions often proved to be. Core memory uses toroids, or rings, of a hard magnetic material, usually a semi-hard ferrite, with each core storing one bit of information. Consider that for a moment. Before this innovation, around 1950, memory was implemented by mercury and nickel-wire delay lines, magnetic drums and Williams tubes. These technologies were temperamental, unreliable, and often required entire rooms to house.
Two key inventions led to the development of magnetic core memory in 1951. The first, An Wang's, was the write-after-read cycle, which solved the problem of how to use a storage medium in which the act of reading erased the data read. This was the foundational puzzle. How do you read something without destroying it in the process? The second, Forrester's, was the coincident-current system, which enabled a small number of wires to control a large number of cores enabling 3D memory arrays of several million bits.
The elegance here is breathtaking. The core to be assigned a value is selected by powering one X and one Y wire to half of the required current, such that only the single core at the intersection is written. Depending on the direction of the currents, the core will pick up a clockwise or counterclockwise magnetic field, storing a 1 or 0. This technique resembles finding a specific apartment in a city by knowing only the street name and the floor number.
What made this technology particularly remarkable was its persistence. Core memory contents are retained even when the memory system is powered down, making it non-volatile memory. In an era when losing power meant losing everything, this property was nothing short of revolutionary.
The Weaving of Digital Tapestries: Manufacturing the Impossible
Here is something many people overlook when discussing early computing: the manufacturing challenge was staggering. Forrester's coincident-current system required one of the wires to be run at 45 degrees to the cores, which proved impossible to wire by machine, so that core arrays had to be assembled by workers with fine motor control under microscopes. Initially, garment workers were used.
Picture this: The speed offered by magnetic-core memory meant that computing could be done in real-time, but manufacturing it was difficult. It was mostly carried out by hand, often by women, who needed microscopes and steady hands to thread thin wires through the tiny holes in the rings. These were not engineers in white coats but skilled craftspeople creating the nervous system of early computers, one microscopic ring at a time.
In 1976, 95% of all computer main memories consisted of ferrite cores, with 20 to 30 billion produced yearly worldwide. The price per bit of core memory was 20 cents in 1960 and decreased from there with 19% per year. By any measure, core memory dominated the computing landscape for decades. It became the most popular form of random-access memory in mainframe and minicomputers from around 1955 to the mid 1970s when it was supplanted by semiconductor memory.
Even today, the term persists. The process of copying the entire content of a computer's main memory to a disk file for further inspection by a system programmer is still called a "core dump." Language, as always, preserves the ghosts of technologies past.
The First Hard Drive: A Room Full of Spinning Platters
While core memory handled the urgent, immediate needs of computation, a different challenge loomed: mass storage. How do you store not kilobytes but megabytes, and access them quickly? In 1954 Reynold Johnson assembled a team at the IBM R&D Laboratory at 99 Notre Dame Ave., San Jose, California charged with developing fast mass storage systems to replace punched cards and magnetic tape in accounting and inventory control applications. Informed by Jacob Rabinow's ideas at NBS, IBM developed and shipped the first commercial Hard Disk Drive, the Model 350 disk storage unit, to Zellerbach Paper, San Francisco in June 1956.
The specifications of this pioneering device seem almost comical by modern standards. On the Model 350, fifty 24-inch diameter disks stacked on a spindle that rotated at 1200 rpm stored 5 million, 6-bit characters, equivalent to 3.75 megabytes of data storage capacity. Today, that amount of data would not hold a single photograph from a modern smartphone.
The team tried rods, strips, tapes and flat plates before settling on an approach that involved magnetizing aluminum disks by coating them with iron oxide paint. Magnetic spots on each disk represented characters of data, and a magnetic arm would read the spots as the disks rotated at blinding speed.
The physical scale was remarkable. The 5 ft high by 6 ft wide unit weighed over one ton, including a separate air compressor required for operation, and leased for $750 per month. That monthly cost, adjusted for inflation, would be thousands of dollars today for what amounts to roughly four seconds of streaming video.
But here is what truly matters: Before RAMAC, information retrieval through a computer took hours or even days. RAMAC could access and manipulate data exponentially faster, in seconds. Speed, as it turns out, changes everything.
Standing Bits on Their Heads: The Perpendicular Revolution
For decades, magnetic recording worked by laying bits horizontally across the disk surface, like dominos placed flat on a table. This longitudinal approach worked well enough, until physics intervened. Hard disk technology with longitudinal recording has an estimated limit of 100 to 200 gigabit per square inch due to the superparamagnetic effect.
What is this superparamagnetic effect? Simply put, as magnetic grains shrink, thermal fluctuations at room temperature can randomly flip their magnetic orientation, erasing data spontaneously. It is like trying to stack coins smaller and smaller until air currents knock them over. Nature, as always, has its limits.
The solution came from Japan. Perpendicular recording was first proven advantageous in 1976 by Shun-ichi Iwasaki, then professor of the Tohoku University in Japan, and first commercially implemented in 2005. The idea was elegantly simple: stand the bits upright rather than laying them flat. Perpendicular recording can deliver more than three times the storage density of traditional longitudinal recording.
Recording by PMR was published in 1977 in the IEEE Transactions of Magnetics. After 28 years of intensive works, perpendicular magnetic recording was commercialized on hard disk drive in 2005. It achieved great success used for all hard disk drives within a couple of years.
Why did it take nearly three decades to commercialize? Because engineering reality rarely matches theoretical promise. The bit size supported by a magnetic medium is inversely proportional to its coercivity and coercivity is limited by the magnetic field achievable by the write element of the recording head. The combination of optimized PMR media and head approximately doubles the magnetic write field achievable with longitudinal recording. A stronger write field enables the use of higher coercivity media with smaller bit sizes and higher areal density.
Toshiba, Seagate and HGST led the commercialization of PMR disk drives between late 2005 and mid-2006. Within two years, in January 2007 Hitachi announced the first 1-terabyte hard drive using the technology. From gigabytes to terabytes, all by standing the magnets upright.
Heat, Light, and the Limits of the Possible: HAMR Technology
Even perpendicular recording has its ceiling. Enter HAMR: Heat-Assisted Magnetic Recording, pronounced like the tool. Heat-assisted magnetic recording is a magnetic storage technology for greatly increasing the amount of data that can be stored on a magnetic device such as a hard disk drive by temporarily heating the disk material during writing, which makes it much more receptive to magnetic effects and allows writing to much smaller regions.
The physics here is fascinating. To increase hard drive capacity, engineers try to fit more data bits, or grains, onto each disk platter, increasing the density of bits crammed into each square inch of surface space. More bits on a disk means more data can be stored. But when bit density is increased, the grains are closer together, so close that the magnetism of each grain can affect the magnetic direction of the grains near it.
The solution involves something remarkable: A small laser diode attached to each recording head momentarily heats a tiny spot on the disk, which enables the recording head to flip the magnetic polarity of a single bit at a time, enabling data to be written. Each bit is heated and cools down in a nanosecond, so the HAMR laser has no impact at all on drive temperature, or on the temperature, stability, or reliability of the media overall.
The solution involves using a laser to heat a single magnetic grain near 450°C. Crucially, this heat is applied only during the write process. The heating and cooling is near instantaneous, taking only a nanosecond.
This technology was far from easy to achieve. The technology was initially seen as extremely difficult to achieve, with doubts expressed about its feasibility in 2013. The regions being written must be heated in a tiny area, small enough that diffraction prevents the use of normal laser focused heating, and requires a heating, writing and cooling cycle of less than 1 nanosecond.
The commercial journey has been long. In the late 1990s, Seagate commenced research and development related to modern HAMR drives. In January 2024, Seagate launched Mozaic 3+, a series of hard drives utilizing HAMR technology, with 28 TB and 30 TB variants.
The key benefits extend beyond raw capacity:
- Heat-assisted magnetic recording is a breakthrough hard drive technology that boosts areal density to enable higher storage capacities, lower power consumption, and reduced total cost of ownership.
- HAMR's higher storage density means fewer physical resources are needed, and its improved efficiency translates to lower power requirements. Seagate's HAMR-based Mozaic 3+ platform promises 2.6× better power efficiency per terabyte.
- HAMR drives have the same form factor as existing traditional hard drives, and do not require any change to the computer or other device in which they are installed.
The Road Ahead: What Comes After HAMR
PMR has reached its superparamagnetic limit at 3TB per platter, where further increases in areal density are constrained by thermal instability. HAMR pushes past this barrier, but researchers are already looking further. HAMR's planned successor, known as heated-dot magnetic recording or bit-pattern recording, is also under development, although not expected to be available until at least 2025.
While Western Digital does not currently sell HAMR drives, the technology is still an integral part of the firm's capacity roadmap. The firm anticipates 36-44TB HAMR drives by 2026, and high-volume shipments of HAMR in 2027. Initial models will be a 36TB CMR drive, a 40TB drive using SMR, and a 44TB UltraSMR drive. By 2030, the firm hopes to have 80TB CMR drives and 100TB UltraSMR drives.
Researchers are already working toward 10TB per platter. This scalability ensures businesses can keep up with increasing data volumes, driven significantly by AI and its data-intensive applications.
Reflections on Seven Decades of Magnetic Memory
From hand-threaded ferrite rings to laser-heated magnetic grains, the evolution of magnetic storage reveals something profound about technological progress. It rarely proceeds in straight lines. Perpendicular recording waited 28 years between invention and commercialization. HAMR took over two decades from concept to consumer product. Patience, it seems, is as essential as brilliance.
What strikes me most is the continuity of purpose. Whether threading wires through tiny donuts in 1955 or firing lasers at spinning platters in 2024, engineers have pursued the same essential goal: store more data, access it faster, keep it safe. The methods change drastically; the mission remains constant.
We now carry more storage capacity in our pockets than existed on Earth in 1960. The data centers powering our connected world hold more information than all the libraries in human history combined. And still, we hunger for more.
The next time you save a file, stream a video, or back up your photographs, consider the invisible architecture making it possible. Somewhere, magnetic grains are flipping their orientations billions of times per second, heated by lasers for nanoseconds, cooled instantly, preserving your memories and your work against the entropy of time. It began with tiny ferrite rings in a laboratory at MIT, and it continues today in the data centers that form the nervous system of modern civilization. The story is far from over.