Samsung and SK Hynix: How Korea Became a Memory Chip Powerhouse

In the age of generative AI, memory chips have become some of the most important components in the global technology industry. AI servers need enormous amounts of DRAM and high-bandwidth memory, or HBM, to move data quickly between processors, and two South Korean companies — Samsung Electronics and SK hynix — sit near the center of that market.

In the second quarter of 2026, Counterpoint Research estimated Samsung’s share of global DRAM revenue at 39% and SK hynix’s at 26%. Together, the two Korean companies accounted for roughly two-thirds of the global DRAM market, although their positions within individual segments such as conventional DRAM and HBM are different.

It is tempting to explain Korea’s semiconductor success with one dramatic story. Korea built an early computer network, Koreans embraced PCs and gaming, PC bangs appeared everywhere, and that enormous domestic appetite for technology somehow produced Samsung and SK hynix.

The real history is not that simple. Samsung had already become the world’s leading DRAM supplier before Korea’s PC bang boom or household broadband revolution had even begun.

What I find more interesting is that several different parts of Korea’s digital history developed during roughly the same decades. While Korean engineers and companies were building networks and semiconductor factories, ordinary people like me were discovering personal computers, assembling our own PCs, connecting to online communities through telephone lines, and eventually experiencing the arrival of an always-connected digital society.

1982 and 1983: Different Parts of Korea’s Digital Future Were Taking Shape

One important date in Korean internet history is May 1982. A research team led by Kilnam Chon connected Seoul National University with the Korea Institute of Electronics Technology, or KIET, in Gumi through a TCP/IP-based network called SDN, or System Development Network.

The connection initially ran at only 1,200 bps, almost unimaginably slow by modern standards. ETRI, the successor institution to KIET, describes the project as Korea’s first TCP/IP network and one of the earliest such networks outside the United States and Europe. You can read more about the project in ETRI’s history of Korea’s early internet.

Something else was happening at almost exactly the same time. In 1983, Samsung made its major push into VLSI and DRAM production, while Hyundai Electronics — the company that would eventually become SK hynix — was also established and entered the semiconductor business.

I do not think one of these developments caused the other. Chon’s network did not create Samsung’s memory business, and Samsung did not build its DRAM operation because Koreans were suddenly using an experimental computer network.

What the dates show instead is that the early 1980s were an important turning point. Korea was beginning to invest seriously in computing, telecommunications, networking, and semiconductor manufacturing at almost the same moment, even though those fields were developing along different paths.

---

How Samsung Reached the Top of the DRAM Market

Samsung’s involvement in semiconductors began before 1983. The company acquired Hankook Semiconductor in 1974, but its major commitment to memory came in the early 1980s, when it chose DRAM as a strategic business despite a large technology gap with established American and Japanese producers.

In 1983, Samsung developed a 64Kb DRAM and began building its Giheung semiconductor facilities. It moved rapidly through successive generations of memory, developing 256Kb and 1Mb DRAM and continuing to invest heavily in production capacity and manufacturing technology.

By 1992, Samsung had reached the world’s top share of the DRAM market. Its official semiconductor history traces that progression from the early 64Kb DRAM period through its rise to global leadership during the 1990s.

The timing matters because Korea’s famous PC bang boom came several years later. Commercial broadband and StarCraft-driven gaming culture exploded near the end of the 1990s, by which point Samsung was already a global DRAM leader.

So Korean gamers did not make Samsung the world’s leading memory company. Samsung’s rise depended much more on long-term capital investment, manufacturing scale, process technology, rapid movement to new memory generations, and a willingness to keep competing through the notoriously cyclical memory market.

---

From Hyundai Electronics to SK Hynix

The history of SK hynix followed a more difficult path. Its predecessor, Hyundai Electronics, was established in February 1983 and began its semiconductor business that same year, completing facilities in Icheon and moving into memory production soon afterward.

In 1985, the company began mass production of 64Kb DRAM. Over the following years it expanded its memory business, and in 1999 Hyundai Electronics acquired LG Semiconductor as part of the major corporate restructuring that followed Korea’s 1997–98 financial crisis.

The enlarged company faced serious financial problems when memory prices collapsed. Hyundai Electronics became Hynix Semiconductor in 2001, survived years of restructuring and financial pressure, and was eventually acquired by SK Group in 2012, becoming the company now known as SK hynix.

The company’s official history and company overview provides a useful timeline from Hyundai Electronics in 1983 through its later development as SK hynix.

The AI boom has given the company another major turning point. SK hynix became particularly strong in HBM, where multiple DRAM layers are vertically stacked to provide the very high bandwidth needed by modern GPUs and AI accelerators.

Samsung and SK hynix therefore arrived at their present positions by different routes. Samsung became the global DRAM leader much earlier, while SK hynix endured restructuring and ownership changes before emerging as a particularly important supplier during the HBM era.

---

The Korea I Remember: Apple II, MSX, 486 PCs and Building My Own Computers

While those companies were learning how to manufacture memory at a global scale, computers were also slowly entering my own life.

My earliest memory of learning about computers goes back to elementary school, probably in the early 1980s. It was such a long time ago that I cannot confidently remember every technical detail, but I remember learning command-line basics — probably DOS — and what I believe was BASIC.

There was also a university professor living in the apartment below ours, and his child was the same age as me. I often went downstairs to play, and their family had an Apple II computer. At that age, I was not thinking about the historical importance of personal computing. I was mostly interested in playing games with my friend, but looking back, that Apple II is one of my earliest memories of using a personal computer in someone’s home.

My first computer at home was an MSX machine. Again, I probably used it much more as a game machine than as an educational computer, which was hardly unusual for a child discovering computers at the time.

By the mid-1990s I was using a 486, based on Intel’s 80486 generation of processors. After entering university, I stopped thinking of a computer as something that had to arrive fully assembled from a manufacturer. I began buying the parts myself and building my own PCs. I chose the CPU, motherboard, RAM, graphics card, sound card, storage, and other components, comparing prices and deciding which part was worth upgrading within my budget.

This was an era when Korea had a strong DIY PC culture, and shopping for individual components was part of the experience of owning a computer. For someone interested in computers, understanding what was inside the case could be almost as enjoyable as using the machine itself. Memory was especially easy to understand as an upgrade. You did not need to know much about semiconductor manufacturing to notice that adding more RAM could make a computer more useful, and users like me became accustomed to thinking about memory capacity as one of the first specifications to check when building or upgrading a PC.

---

When Going Online Meant Taking Over the Family Telephone

The next major change in my computer life came with PC communication services. Once I connected a modem, the computer stopped being a machine I used entirely by myself and became a way to communicate with people elsewhere. There was one major problem: the modem used the household telephone line. When I was connected to a PC communication service, our home phone was effectively busy.

That is difficult to imagine now, when almost everyone carries an individual smartphone. At the time, however, one landline was often the main telephone for the entire family, so occupying it for hours was not a minor inconvenience. People trying to call our home would hear a busy signal, and my parents sometimes became angry because the phone seemed to be permanently unavailable. The more fascinated I became with being online, the more frustrating the situation became for everyone else in the house.

The telephone bill was an even bigger problem. I used PC communication services so much that our household bill sometimes reached around KRW 300,000 to 500,000 a month. That figure is my own family’s experience, not a claim that every Korean PC communication user paid that much. I was simply spending an extraordinary amount of time connected, and in an era when online time could translate directly into telephone charges, enthusiasm could become expensive very quickly. Eventually, our family began using CO-LAN.

CO-LAN, short for Korea Telecom’s public corporate communications network, was originally designed primarily as a data-network service. For home users, it could use a VDM device to separate voice and data frequencies on the existing telephone line, which meant that the telephone and computer communication could be used at the same time. A contemporary 1996 report described exactly the family problem I remember: children spending long periods on PC communication, parents being unable to use the home telephone, and worries about mounting telephone charges. CO-LAN offered a way to use voice and data simultaneously without installing an entirely separate physical telephone line.

When I used it, the biggest difference was the feeling of freedom. Instead of making a normal modem call every time and constantly thinking about how long I had been connected, getting onto PC communication felt almost immediate, and my family could still use the telephone. Compared with today’s fiber connections, Wi-Fi, and 5G, the speed was extremely slow. But to someone who had experienced dial-up bills and constant complaints that the home phone was busy, it felt like a huge technological change.

That period came just before Korea’s broadband explosion. Within only a few years, the idea of paying attention to every minute spent online would begin to disappear as flat-rate, high-speed household internet became increasingly normal.

---

Broadband and PC Bangs Created a Digital Society, Not the Memory Giants

By the late 1990s, Korea’s relationship with computers was changing at remarkable speed. Broadband networks expanded, multiplayer games became much easier to access, and PC bangs began appearing in neighborhoods around the country.

StarCraft, released in 1998, became closely connected with this transformation. PC bangs gave people fast connections and capable computers without requiring every young person to own a high-end machine at home, and gaming increasingly became a social activity as well as an individual one. I have written much more about that period, including my own memories of StarCraft and how PC bangs changed over time, in my article on Korean PC bang culture.

But this is also where I want to separate two stories that are sometimes too easily combined. PC bangs, broadband, gamers, and rapidly increasing PC ownership certainly increased Korea’s appetite for digital technology, but they did not create Samsung’s or Hyundai Electronics’ semiconductor competitiveness. Samsung was already the world’s leading DRAM supplier by 1992. Hyundai Electronics was already producing memory years before Korea became famous for PC bangs.

The relationship is therefore more interesting as parallel development than simple cause and effect. Korean companies were becoming globally competitive manufacturers at the same time that Korean households were becoming unusually enthusiastic users of computers and communication networks.

That digital environment later supported many other industries as well. Korea’s early broadband culture helped create the conditions in which online games, portals, digital entertainment, and eventually mobile-first formats could develop rapidly.

---

From DRAM to HBM: Why Memory Matters Even More in the AI Era

The technology has changed dramatically since my 486 days, but memory has become even more important. Traditional PCs needed DRAM so that programs and operating systems could work quickly. AI systems now require vast amounts of memory bandwidth to move enormous datasets between processors, which has turned HBM into one of the most strategically important semiconductor products in the world.

In the second quarter of 2026, Samsung held about 39% of global DRAM revenue and SK hynix about 26%, according to Counterpoint Research. Their combined share shows why discussions about global memory supply so often lead back to South Korea, although Micron and rapidly growing Chinese manufacturers mean the competition is far from settled.

The AI era has also changed the balance between the two Korean companies. Samsung remains exceptionally strong across the broad DRAM market, while SK hynix has established a particularly strong position in HBM, where its early relationships with major AI customers gave it an important advantage. Memory is also once again having an enormous impact on Korea’s exports. In July 2026, South Korea’s semiconductor exports reached $41.01 billion, up 178.8% from a year earlier, as demand related to AI servers, memory, and enterprise computing remained exceptionally strong.

When I see numbers like these, I sometimes think back to the computers I knew when I was young. I remember entering commands at school, playing games on the Apple II downstairs, using an MSX at home, moving to a 486, assembling my own PCs, and staying online so long that my parents could not use the telephone.

None of those things made Samsung or SK hynix successful. What happened is subtler and, to me, more interesting. During the same decades in which ordinary Koreans were rapidly learning to use computers and networks, Korean engineers and companies were learning how to manufacture memory at a scale and technological level that could compete with the world’s largest semiconductor companies. The two histories grew alongside each other.

I now see Korea’s technology story as a much longer history of infrastructure, manufacturing, investment, research, corporate survival, and people adapting to new generations of digital technology.

From a 1,200-bps research network and 64Kb DRAM to HBM for AI servers, the scale has changed almost beyond recognition. Yet for someone who lived through part of that transition, the history of Samsung and SK hynix is also connected to something much more ordinary: the memory of a computer, a telephone line, and the excitement of discovering that a machine in your home could suddenly connect you to a much larger world.

---