Korea's Corporate Giants at a Crossroads: Capital, AI, and the End of Business as Usual
The word chaebol carries a specific weight in global business. It describes something that does not translate cleanly: a family-controlled industrial conglomerate with fingers in semiconductors, shipbuilding, insurance, construction, retail, and entertainment simultaneously — bound together not by pure market logic but by a web of cross-shareholdings, founding-family governance, and decades of state-aligned industrial policy. For most of the twentieth century, that model worked. It built South Korea's economy from a postwar baseline to a global manufacturing powerhouse in less than two generations. What it was never designed to do was navigate an AI-driven technological disruption at the speed and scale the current moment demands. That is exactly the challenge the chaebol face in 2026, and the most consequential among them are not waiting for the pressure to ease. They are reinventing themselves — and doing it with a ferocity that should get the attention of every serious investor watching Asia.
![]() |
| Seoul's corporate skyline reflects a new economic ambition: an AI-driven economy targeting G7-level status by 2027. |
The reinvention is not cosmetic. Samsung Electronics has committed a record $73 billion in semiconductor capital expenditure and research for 2026 alone — the largest single-year semiconductor investment by any company in history, representing a 22 percent increase over its already substantial 2025 spending. SK Hynix, which holds approximately 53 percent of the global high-bandwidth memory market, is expanding its infrastructure investment by more than four times its previously announced figure, constructing new fabrication facilities that will extend its supply capacity through 2027 and beyond. Hyundai Motor Group debuted the commercial-ready Atlas humanoid robot at CES 2026 and laid out a formal Physical AI strategy that positions the group as a global leader in robotics manufacturing. These are not incremental adjustments. They are structural bets placed on a single thesis: that AI infrastructure demand will define the next decade of global corporate competition, and that Korea's conglomerates intend to be on the supply side of it.
Samsung: The $73 Billion Reset
Samsung's 2026 investment plan arrives after a period of uncommon vulnerability for the company. Its HBM3E qualification process was troubled throughout much of 2025, with the 12-layer stack failing to meet Nvidia's full performance standards for the premium global market. SK Hynix and Micron captured the dominant share of those supply contracts, and Samsung found itself in the uncomfortable position of being the world's largest memory chip producer while watching a rival supply the most commercially significant segment of the market. The company's response was not defensive consolidation. It was a full-scale technological reset centered on the sixth-generation HBM4.
Samsung sent HBM4 samples to Nvidia in September 2025, entered final qualification in late 2025, and began mass production in February 2026 at its Pyeongtaek campus. By the time the qualification was confirmed, the market reaction was swift: Samsung had reportedly sold out its entire 2026 HBM4 production capacity before mass production formally commenced. Nvidia CEO Jensen Huang visited Samsung's booth at GTC 2026 in San Jose and publicly described the relationship as a great partnership, adding that Samsung was handling production for Grok 3, the next-generation inference processor built on technology from Nvidia's Groq acquisition. OpenAI and Broadcom selected Samsung as an additional HBM4 partner for OpenAI's custom Titan processor. AMD expanded its supply partnership to include HBM4 for its Instinct MI455X accelerators. In the span of roughly six months, Samsung moved from qualified-for-secondary-markets to supplier to every major AI hardware ecosystem simultaneously.
![]() |
| Memory semiconductors remain the engine of South Korea's export growth, with HBM demand surging alongside the global AI infrastructure boom. |
The $73 billion investment addresses three simultaneous objectives. The first is scaling HBM4 production capacity aggressively enough to meet a demand curve that has caught the entire memory industry off guard. Samsung is targeting a 50 percent expansion in overall production capacity by late 2026, with monthly 1c DRAM capacity scaling to 200,000 wafers by year end. The second objective is advancing its foundry business — currently holding approximately 7.2 percent market share against TSMC's dominant 69.9 percent — through 2nm process node development and strategic customer relationships, including a multi-billion-dollar deal to manufacture AI chips for Tesla's autonomous driving systems. The third is positioning Samsung's balance sheet for a sustained multi-year investment cycle, backed by what Morgan Stanley projects could be a greater than 150 percent increase in earnings per share for 2026. The rational underlying logic is straightforward: AI infrastructure capital expenditure from Microsoft, Google, Meta, Amazon, and others is locked in for years. Samsung intends to be structurally embedded in that spending before any competitor can displace it.
SK Hynix: Defending the HBM Crown
SK Hynix occupies a different strategic position than Samsung, and its 2026 priorities reflect that difference clearly. As the incumbent HBM market leader — the company that secured the dominant share of Nvidia's HBM3E supply contracts and established the foundational supply relationship with the leading AI accelerator maker — SK Hynix's primary challenge in 2026 is not market entry. It is capacity expansion fast enough to prevent its customers from having to diversify away, combined with a technology roadmap that keeps it ahead of Samsung's now-accelerating HBM4 push.
The company is constructing its M15X fabrication facility, scheduled for utilization by mid-2027, and its Yongin Cluster 1 plant is projected for completion in the second quarter of 2027. SK Group, the parent conglomerate, also announced an Nvidia-powered AI factory to support semiconductor research, development, and production — a signal that the group intends to embed AI capabilities directly into its own manufacturing infrastructure rather than simply producing components for others' systems. In October 2025, Samsung and SK Hynix joined forces on a separate front, signing a letter of intent with OpenAI for the eventual supply of 900,000 DRAM wafers per month — a collaboration that speaks to the scale of AI memory demand and the recognition, even among domestic rivals, that certain supply challenges require coordinated responses.
For foreign investors benchmarking SK Hynix, the key data point is market structure rather than individual company metrics. Samsung and SK Hynix together controlled approximately 90 percent of global HBM supply as of 2025. That duopoly, operating in a market where global memory semiconductor sales growth was revised upward to 39.4 percent for 2026 by the World Semiconductor Trade Statistics organization, represents a supply chain position with few parallels in any technology sector. The risk is not demand — it is execution: sustaining yield rates, maintaining qualification with leading AI chipmakers, and building out capacity fast enough to meet a demand curve that has surprised even the most optimistic industry forecasts.
Hyundai and Physical AI: The Robot Is Now a Product
Hyundai Motor Group's pivot to AI robotics is the most strategically complex of the major chaebol transformations, because it requires the group to simultaneously defend its core automotive business — where the shift to electric vehicles and software-defined vehicles is reshaping competitive dynamics — while making a credible play for leadership in an entirely new industry category. The CES 2026 unveiling of its AI Robotics Strategy under the theme "Partnering Human Progress" was more than a marketing presentation. It was a structural commitment, anchored by the public commercial debut of the next-generation Atlas humanoid robot and a formal partnership with Google DeepMind to accelerate development of next-generation humanoids.
The operational details matter. Atlas is scheduled for deployment at Hyundai Motor Group's Georgia manufacturing facility by 2028, initially performing sequencing tasks that reduce physical burden on human workers. A new robotics production facility with annual capacity of 30,000 units has been announced, positioned as a key hub for the growing global robotics ecosystem. The End-to-End AI Robotics value chain spans Hyundai Motor, Kia, Hyundai Mobis, and Hyundai Glovis — meaning that the group's existing manufacturing scale becomes both the training ground and the first commercial deployment environment for the robotics business simultaneously. The Robot Metaplant Application Center and Software-Defined Factory platform provide the data infrastructure for this deployment, turning Hyundai's own production lines into proprietary training assets for AI robotics systems.
![]() |
| Rising real wages and expanded fiscal support are bringing South Korean consumers back into the market, a structural shift that matters well beyond the GDP headline. |
The governance dimension of Hyundai's robotics push adds a layer of complexity that is specifically relevant to investors with an equity perspective. Boston Dynamics, the US robotics unit acquired from SoftBank in 2021 for $660 million at a 60 percent group stake, is being prepared for a potential Nasdaq IPO as early as 2027. Executive Chair Chung Euisun holds a personal 20 percent stake in Boston Dynamics, acquired for roughly $220 million, which current brokerage estimates value at $13.6 billion or more — based on valuations ranging from 100 trillion to 150 trillion won applied by KB Securities and Hanwha Investment and Securities respectively. A successful listing would provide Chung a capital resource of that scale, with analysts suggesting it would ease the inheritance tax burden that has historically complicated chaebol succession and simultaneously create the financial flexibility to restructure Hyundai's famously complex circular ownership arrangement. Hyundai remains the only major Korean conglomerate that still operates within a circular shareholding structure — a governance configuration that has drawn sustained institutional investor criticism and regulatory scrutiny for years.
Corporate Governance: The Reform Variable That Changes Everything
For American institutional investors, the governance question is often the decisive variable in assessing Korean equity exposure. The chaebol ownership model — concentrated family control, cross-shareholdings that dilute minority shareholder influence, and historically opaque related-party transactions — has been a persistent discount applied to Korean stocks relative to global peers with comparable earnings profiles. That discount, sometimes referred to as the "Korea discount," has been estimated at anywhere from 20 to 40 percent relative to comparable firms in other markets.
The current reform dynamic has multiple catalysts converging simultaneously. The government under President Lee has maintained pressure on corporate governance improvement, building on the momentum of the prior administration's Corporate Value Enhancement Program. Samsung Electronics disclosed its 2026 Corporate Value Enhancement Plan, which includes a commitment to regular dividends of 9.8 trillion won and outlines shareholder return mechanisms alongside its investment program — a deliberate signal that record capital expenditure and shareholder value are not mutually exclusive priorities. The AI investment cycle itself creates pressure in the same direction: when companies need sustained access to global capital markets to fund multi-year, multi-trillion-won programs, the governance standards demanded by international institutional investors carry real consequences for cost of capital.
The professional management dimension is evolving in parallel. Samsung's November 2025 organizational restructuring — dissolving the HBM special development team and integrating its personnel into the core DRAM development department under a unified Memory Technology Development Department — reflects a move toward cleaner, more accountable technology leadership rather than the parallel-command structures that complicated its HBM3E development cycle. The appointment of specialists with specific technical mandates to senior positions, rather than generalist family-aligned executives, is a trend visible across the major chaebol groups, even where it has not yet reached the level of formal governance reform.
The Investment Case: What Chaebol 2.0 Means for Foreign Capital
The convergence of AI investment momentum and governance reform creates a specific opportunity window for investors who have historically avoided Korean equities on governance grounds while acknowledging the underlying business quality. The Chaebol 2.0 thesis is not that family control is disappearing — it is that the families running these groups understand that sustaining their positions in the AI era requires the kind of structural credibility and capital market access that opaque governance structures directly undermine. Chung Euisun's personal financial stake in a Boston Dynamics IPO is a more powerful governance alignment mechanism than any regulatory mandate could produce.
For US investors specifically, the supply chain positioning argument is straightforward. Every major American AI infrastructure company — Nvidia, AMD, Microsoft, Google, Amazon, Meta, OpenAI — either has existing supply contracts with Samsung or SK Hynix for HBM, or is actively pursuing them. That dependency is structural and medium-term in duration. The memory capacity being built in Korea through 2027 and beyond is the physical substrate on which American AI ambitions will run. Hyundai's robotics play adds an additional dimension: the Physical AI category — robots, smart factories, autonomous systems that learn from real-world data — is the next major deployment frontier after cloud AI infrastructure, and Hyundai has positioned its group-wide manufacturing network as both the development environment and the initial market for that category.
The chaebol are not finished products. They are works in progress — heavily capitalized, strategically aggressive, and operating in categories where demand is structurally locked in for years. The transformation from family legacy to AI supply chain anchor is messier and slower than any investment deck would suggest, and the governance risks are real rather than theoretical. But the direction of travel is clear, and the companies moving fastest are the ones with the most to prove after difficult years. Which chaebol's AI bet do you think carries the highest upside — and the highest execution risk — for foreign investors entering the Korean market now?
Continue your journey into Korean life below:
- culture / insight / k-culture / pillarMar 15, 2026
- culture / insight / k-culture / pillarMar 15, 2026
- k-culture / ktoday / living / pillarMar 15, 2026
.webp)
.webp)

.webp)
.webp)
.webp)
0 Comments