SK Hynix employees have become the darlings of blind dates, while Samsung workers have been on strike for eighteen days—same issue, two different answers.

The AI memory supercycle has propelled two South Korean companies to the pinnacle of profitability. Yet this same boom has unfolded differently: at SK Hynix, it sparked a wealth‑sharing experiment that involved the entire workforce, while at Samsung, it ignited a strike that could at any moment deal a severe blow to the global AI supply chain. This divergence goes beyond mere differences in labor‑management practices; it reflects how two distinct governance logics fracture under identical extreme pressures—and reveals a larger risk that neither side has yet recognized.

I. SK Hynix: Institutionalized allocation is a strategy, not a welfare benefit.

SK Hynix's most widely reported headline lately has been a story from the dating‑and‑marriage market: South Korean media, citing matchmaking agencies, reported that SK Hynix employees receive an "unconditional A‑grade rating" in blind‑date matches, while variety shows have even begun poking fun at the "market value" of the company's work uniforms. The absurdity of this detail perfectly captures the intensity of contemporary South Korean society's sentiments about AI‑driven wealth distribution.

The figures themselves are already staggering. In February 2026, SK Hynix paid out its FY2025 performance bonuses: the amount equaled 2,964% of an employee's monthly base salary, meaning a worker earning an annual salary of 100 million won received an additional roughly 148 million won; the total bonus pool amounted to approximately 4.5 trillion won. And this is just the beginning—based on analysts' projections of around 250 trillion won in operating profit for 2026, the average bonus per employee among the company's 35,000 staff is expected to be about $477,000, rising to nearly $900,000 by 2027.

But what truly deserves attention is not the magnitude of the numbers, but the structure of the mechanism.

In September 2025, SK Hynix reached an agreement with its labor union to abolish the previous 1,000% cap on bonuses and instead allocate 10% of annual operating profit—without any upper limit—to all employees. Both parties also agreed to maintain this framework for a period of ten years. This is not a one-time gesture of generosity but rather a long-term commitment that ties employees' interests to the company's future at the institutional level.

Management's logic is not a purely moral choice, but rather a sober strategic calculation: in the AI era, HBM engineers are truly scarce production resources. SK Hynix currently dominates the HBM market with a 62% share, and NVIDIA sources roughly 90% of its HBM needs from SK Hynix. At such a high level of concentration, the loss of even a single batch of core engineers represents an immediate competitive disadvantage. Institutional commitments are more effective at retaining talent than one-time allocations—this is not a benefit; it is the cost of maintaining a competitive moat.

This logic has already exerted a discernible downward pressure on the talent market: admission scores for related university majors have surged, approaching the thresholds of medical schools—reflecting the labor market's collective amplification of these signals.

II. Samsung: Why, despite identical demands, the situation here has reached the brink of a strike?

Current situation

On April 23, approximately 40,000 members of the Samsung Electronics union gathered at the Pyeongtaek plant—the world's largest semiconductor manufacturing facility—to formally announce their strike plan. The union's three core demands are: abolishing the current 50% cap on bonuses, enshrining a 15% operating‑profit‑based bonus in the labor contract as a permanent provision, and securing a 7% wage increase.

During the negotiations, the two sides at one point came close to reaching a consensus on a 13% profit-sharing arrangement—equivalent to roughly $340,000 per employee. However, their differences remained locked on a single issue: management was willing to offer only a one-time payout, while the union insisted that it be enshrined in the contract and implemented annually. On May 12, the final round of government‑mediated talks collapsed, sending Samsung's stock down as much as 6.1% that day. A 18‑day general strike is scheduled to begin on May 21.

Two structural factors explain why Samsung's management has been unable to institutionalize its practices.

Reason one: The cost of institutionalizing this arrangement far outweighs the sum itself. SK Hynix is a pure‑play memory company with a highly concentrated profit base; the 10% wage increase was already factored into its cost structure when the agreement was signed. By contrast, Samsung Electronics is a diversified conglomerate, with semiconductors, smartphones, home appliances, display panels, and foundry services all under one roof. If the union's demand for a 15% raise were enshrined in the contract, the resulting bonus pool would swell to roughly KRW 45 trillion—equivalent to SK Hynix's total operating profit for the entire year of 2025. Once such a figure becomes a contractual term, it would serve as a benchmark that every division within the group could invoke—prompting the DX and Display divisions to immediately engage in negotiations based on the same logic. Institutionalizing even a single precedent would entail a systemic overhaul of the entire Samsung Group's compensation framework. What management truly fears is not "being unable to afford it," but rather "locking in a standard that cannot be replicated across the entire group once it has been implemented."

Reason Two: There are fissures within the union as well. Roughly 80% of the members of Samsung's largest union come from the semiconductor division, so the bargaining agenda naturally leans toward chips. Meanwhile, workers in the smartphone and home‑appliance divisions seek to include company‑wide profits in the distribution; more than 2,500 members have already left the union because their demands favor the chip sector. This strike is far from monolithic—it is, in itself, a microcosm of the broader dispute over profit sharing.

Cultural Differences: The Same Roots, Different Soils

Behind management's resistance lies a deeper organizational‑cultural issue—one that also happens to be the direct reason Samsung has lagged behind in the HBM market.

Samsung has long maintained a highly centralized, top-down decision-making culture. Engineers commonly report that proposals without precedent are almost impossible to get approved; under performance‑evaluation pressure—where contracts are renewed only upon delivering results—managers push for short‑term deliverables, leading to ever‑shorter R&D cycles. While this culture once served as a competitive advantage during Samsung's era of dominance in the DRAM market, on a track like HBM—requiring substantial investment in cutting‑edge R&D and organizational agility—it has begun to weigh it down.

By contrast, SK Hynix's corporate culture is renowned for fostering innovation and knowledge sharing, widely regarded as the organizational foundation that enabled it to catch up and excel in the HBM market.

The most direct signal is a reversal in the flow of talent.From December 2025 to the present, approximately 200 Samsung engineers have moved to SK Hynix—a phenomenon that has been extremely rare over the past decade. This shift in compensation structures is quietly but irreversibly reshuffling the semiconductor industry's most critical resource: human capital.

III. The True Cost of the 18-Day Strike

The most direct financial loss from an 18-day shutdown, analysts estimate, would amount to a 7%–12% impact on operating profit; the union's own estimate of KRW 30 trillion, however, is widely regarded as an extreme scenario.

But the real loss lies not in the financial statements, but in the time window.

The current HBM market landscape is clear: SK Hynix leads with a 62% share, followed by Micron at 21%, and Samsung in third place with roughly 17%. Samsung previously missed the first-mover advantage in HBM3E, and the "Samsung is back" signal on the client side only emerged earlier this year. The timeline for mass production of HBM4 will be a critical juncture determining whether Samsung can reclaim its position among the top tier of competitors.

The 18-day strike has occurred at this juncture, and its cost is not a one-time financial loss but rather the systemic risk of delaying Samsung's HBM4 mass-production schedule.The single-day operations at the Pyeongtaek plant have already reduced output per shift by 58%; if mass production is delayed, SK Hynix will gain a longer period of exclusive access, further widening the market‑share gap between the two. In highly concentrated B2B customer relationships, such a gap often exhibits self‑reinforcing inertia.

Samsung's management and the union are disputing the allocation ratio, but in doing so they have inadvertently created a timing gift that benefits their competitors the most.

IV. The negotiator who was not at the table on either side

SK Hynix's ten-year commitment and the Samsung union's demands enshrined in the contract both frame this dispute as a domestic Korean issue—between management and labor, between the company and its employees. Yet there is one factor that neither features at the negotiating table nor is fully priced into the market:Washington。

The historical precedent is strikingly clear. Between 1978 and 1986, Japan's global DRAM market share surged from under 30% to roughly 75%, while the U.S. share plummeted from 70% to 20%. In response, the United States compelled Japan in 1986 to sign the Semiconductor Agreement, which imposed minimum pricing for DRAMs and mandated that foreign‑made chips capture more than 20% of the Japanese market within five years—an overt act of political intervention aimed squarely at the monopolistic super‑profits generated by Japan's then‑nearly 90% market share.

Currently, the concentration of South Korea's two major companies in the HBM sector closely mirrors Japan's situation at the time. About 90% of SK Hynix's HBM shipments go to NVIDIA, and this super‑profit chain has its origins and end points in the United States—leaving the substantial profits along the way in the suburbs of Seoul. According to analysts' forecasts, Samsung and SK Hynix could see combined profits this year approach 500 trillion won, potentially generating corporate tax revenues that exceed the South Korean government's full-year corporate tax revenue target. This figure itself has become a political flashpoint.

The Trump administration has both the motivation and the tools: the security of its allies serves as an immediate bargaining chip, and its policy toolkit is extremely diverse. Its governing style, which prioritizes short-term, visible results, ensures that once attention shifts to this issue, action will be swift.

The timing is a key variable. At present, Trump's attention is fully consumed by the situation in Iran and the midterm elections, and the excess profits in the AI hardware supply chain have not yet made it onto the agenda. However, once the midterm elections are over and the external situation begins to stabilize, this will be a goal with extremely high political returns and very little resistance to implementation.

This, in fact, is the non‑consensus that this article seeks to highlight: two companies and their unions are vying—through contracts and strikes—for the allocation framework of the next decade; yet the precondition for such a framework—the continued retention of these funds in South Korea by Washington—has not been discussed by any party at the negotiating table.

Conclusion

SK Hynix's story is that of a focused company that, at the right juncture in the AI cycle, has aligned employee interests with the company's destiny through thoughtful institutional design, thereby reinforcing its engineering‑centric culture and building a robust talent moat. Samsung, by contrast, is a larger, more complex enterprise that is paying the price for its scale and historical legacy—not only at the negotiating table, but also within its engineering labs.

But the stories of both companies are nested within a larger narrative.In the age of AI, where excess profits are generated, where they are retained, and who gets to decide—these have never been issues confined solely to firms and their employees.The size of the pie they're vying for is subject to third-party opinions.

27K Views