Recently, many people have been asking as they watch Samsung Electronics (005930.KS) and SK hynix (000660.KS) share prices surge sharply in the stock market: "Why on earth are they rising this much?"
As a semiconductor industry professional, the answer I've gathered from conversations with fellow industry CEOs, executives, and AI is simple yet clear. The larger the artificial intelligence (AI) industry grows, the more semiconductors are needed — exponentially so. Unlike when the metaverse was the buzzword, the AI industry clearly has real substance, and many readers will sense that its growth pace is accelerating even further.
AI demands enormous amounts of memory in the process of learning, computing, and storing data. High Bandwidth Memory (HBM), which has recently been in the spotlight, requires far higher performance than conventional DRAM, and its demand is surging relative to supply. Moreover, in last year alone, users created more than 10 billion pieces of content using AI — a more than 100-fold increase from fewer than 100 million pieces per year during YouTube's early days in 2006.
The gap becomes even more stark when measured by memory usage. From tens of petabytes during YouTube's early days, memory usage from AI-generated content reached 181 zettabytes last year — an increase of more than one million times. Within this AI-driven current, Samsung Electronics and SK hynix are being re-evaluated not merely as the world's No. 1 and No. 2 memory suppliers, but as core global players in AI infrastructure. In other words, the recent stock price rally is not mere speculation but a reflection of rapid structural change and growth in the AI industry. Recently, there has been much discussion that memory is the bottleneck in AI industry development, and some have even suggested that memory companies' market capitalizations could surpass those of xPU semiconductor companies. However, it seems difficult to overtake the market cap of a company like Nvidia, which is building not just GPUs but an entire platform, software ecosystem, and a robust partner network.
China's semiconductor pursuit is also formidable. I personally attended "SEMICON Korea 2026" held at COEX in February and "SEMICON China 2026" held in Shanghai last month, comparing the semiconductor trajectories of both countries.
At SEMICON Korea, AI semiconductors, advanced packaging, and high-performance memory technology were the key topics. HBM and 2.5D/3D packaging technologies in particular signaled that the industry's center of gravity is shifting beyond a simple trend. SEMICON China, by contrast, had a different hue. China still showed strong determination toward localizing equipment and materials, focusing on building its overall semiconductor ecosystem beyond just memory. There were also clear moves to expand gradually from low- and mid-range processes into increasingly advanced territory.
Comparing the two exhibitions makes things clear. Korea is pursuing a strategy of technological depth and maintaining a wide lead, while China is pursuing a strategy of speed and scale. This gap will be difficult to close in the short term, but it also demands constant vigilance and innovation from Korea.
The growth trajectory of Samsung Electronics and SK hynix over the coming years is already a foregone conclusion. If the numerous materials, components, and equipment companies that support Korea's semiconductor industry from the second tier can participate more broadly in the growth of these two giants, it is self-evident that our country can pursue wider and more stable growth beyond large-conglomerate-driven development.
The semiconductor industry — as even Nvidia, the world's leading semiconductor company, demonstrates — is no longer a structure that any single company can dominate. In the AI era, ever more precise materials, components, and equipment technologies are required for semiconductor advancement. This represents both an opportunity and a challenge for these suppliers. The key is to move beyond simple vendor relationships. Co-development, early-stage participation, and technology internalization will and must become the core keywords. A structure where memory companies and their suppliers design and validate together from the earliest stages must take root.
Government and the broader industry must also focus not on short-term results but on building a mid- to long-term technology ecosystem. Only then can Korea leap beyond being a memory semiconductor powerhouse to become a nation with a sustainable semiconductor ecosystem.
A single small chip transforms an industry, and that industry transforms a nation's future. In the AI era, Korea has been presented with a great opportunity that may never come again, and we have entered the early phase of an upward curve toward even greater development and change. To sustain semiconductor growth for 10 to 20 years or more, now is the time for government leaders and academia-industry-research leaders to put their heads together, refine strategies and policies, and execute — that thought weighs heavily on my mind these days.
He is…
· Bachelor's degree in Metallurgical Engineering and Master's/Ph.D. in Materials Science and Engineering, Seoul National University
· Principal researcher, Memory Business Division, Samsung Electronics
· Industrial researcher dispatched to IMEC, a Belgian semiconductor research institute
· Senior researcher, Samsung Electronics Semiconductor Research Center
