Before the formal announcement, media reports had described a possible package worth more than 100 trillion won—about $72 billion—funded mainly by cash dividends and based on returning 50% of free cash flow. The reported figure should therefore be distinguished from Samsung’s subsequent company announcement, which gave a 90–110 trillion won estimate and specified the initial dividend and employee-compensation components.
Samsung CFO Park Soon-cheol had said the board and management were actively discussing a special dividend and other parts of the shareholder-return program. At the same time, he emphasized the need to balance distributions with investment, as Samsung expected chip shortages to become more acute and potentially continue through 2028 amid long-term data-center demand.
That balance is central to the plan. Samsung was not presenting shareholder returns as a substitute for investment in high-bandwidth memory and other advanced chips. Rather, the policy aimed to distribute a share of exceptional free cash flow while retaining enough capital to pursue the next phase of AI-memory demand.
SK Hynix’s board approved a 40 trillion won buyback, equivalent to about $28.6 billion. The company planned to repurchase up to approximately 24.07 million shares on the open market from August 20 through November 19, then cancel all of the shares acquired. The amount represented about 3.3% of shares outstanding.
The cancellation matters because it permanently reduces the company’s share count. If earnings remain unchanged, fewer shares divide the same earnings base, potentially increasing earnings per share. The buyback also returned cash directly to shareholders who sold into the program, while the cancellation provided a stronger per-share signal than holding repurchased shares in the treasury.
SK Hynix simultaneously raised its medium-term shareholder-return target. For cumulative free cash flow generated from 2025 through 2027, it moved from returning “within 50%” to returning more than 50% through repurchases, cancellations and dividends. The company also left open the possibility of a special dividend and further measures.
SK Hynix said its intrinsic value—based on its business competitiveness and cash-generation capability—was not fully reflected in its current share price. In practical terms, management was signaling that it viewed the post-selloff valuation as too low and was willing to use its balance sheet to act on that view.
The timing reflected a clash between record AI-related profits and investor concerns that the boom could be approaching a peak. Shareholders wanted a larger share of the cash being generated by the memory cycle, while chip stocks fell as markets questioned whether AI infrastructure spending and memory prices could remain elevated.
SK Hynix had fallen more than 50% over two months before announcing its buyback, according to reporting on the plan. Investors had also focused on the companies’ large expected cash balances and their existing policy of returning roughly half of free cash flow, arguing that more could be distributed after the AI-driven earnings surge.
The response therefore served two purposes:
But the measures did not prove that memory prices would continue rising. A buyback reduces the number of shares and signals management confidence; it does not determine future supply, pricing or demand. Those fundamentals still depend on AI capital expenditure, memory supply discipline and broader financial conditions.
Investors broadly treated SK Hynix’s cancellation as a strong vote of confidence in the company’s near-term business outlook and AI-server demand. The size and speed of the program suggested that management considered the market’s valuation concerns excessive relative to the company’s operating prospects.
The interpretation was more nuanced than “buybacks guarantee higher chip prices.” The announcement could support the stock through a smaller share count and improved confidence, but memory remains a cyclical industry. If AI spending slows or supply expands faster than demand, shareholder returns would not prevent pressure on profits.
Samsung’s plan carried a related message, although its larger emphasis on dividends made it a different type of response. Cash dividends provide a direct payout, while the company’s retention of investment capacity kept the focus on whether it could capture continuing demand for high-bandwidth memory and data-center chips.
The announcements were followed by a sharp rebound in South Korean equities. On August 20, SK Hynix rose 12.73% to 1.691 million won and Samsung Electronics gained 9.49% to 271,000 won. The Kospi advanced 5.9% and moved back above 6,800.
The rally reflected both the shareholder-return news and a broader improvement in risk appetite as U.S. Treasury yields eased. Higher long-term Treasury yields had previously weighed on growth and semiconductor valuations, so the market reaction was not solely a judgment on the companies’ payout policies.
The announcements also helped support a wider semiconductor rebound. Still, analysts continued to watch two unresolved risks: whether high U.S. yields would pressure technology valuations again and whether a future slowdown in AI infrastructure spending would weaken memory demand.
The scale of the plans was made more credible by the companies’ strong cash-generation backdrop. Reuters reported that Samsung and SK Hynix together were expected to hold $263 billion in net cash by year-end.
Micron was also discussed as a comparison after pledging to return 100% of “excess cash” to shareholders over the long run, according to reporting cited in the source material. That is not directly equivalent to Samsung’s 50%-of-free-cash-flow framework or SK Hynix’s more-than-50% cumulative target: “excess cash” and free cash flow are different capital-allocation concepts, and the available reporting does not establish identical definitions or timetables.
The August measures changed the debate around Korea’s memory-chip leaders from whether they would share their AI windfall to how much they could return while maintaining the investment required for the next memory cycle.
Samsung offered a large, dividend-led return with continued reinvestment. SK Hynix delivered a more immediate valuation signal through a 40 trillion won repurchase and cancellation, while lifting its cumulative payout target above 50% of free cash flow. Both approaches supported their stocks in the short term, but neither removed the fundamental test ahead: whether AI infrastructure demand and advanced-memory pricing can justify the earnings expectations embedded in the companies’ valuations.
Before the formal announcement, media reports had described a possible package worth more than 100 trillion won—about $72 billion—funded mainly by cash dividends and based on returning 50% of free cash flow. The reported figure should therefore be distinguished from Samsung’s subsequent company announcement, which gave a 90–110 trillion won estimate and specified the initial dividend and employee-compensation components.
Samsung CFO Park Soon-cheol had said the board and management were actively discussing a special dividend and other parts of the shareholder-return program. At the same time, he emphasized the need to balance distributions with investment, as Samsung expected chip shortages to become more acute and potentially continue through 2028 amid long-term data-center demand.
That balance is central to the plan. Samsung was not presenting shareholder returns as a substitute for investment in high-bandwidth memory and other advanced chips. Rather, the policy aimed to distribute a share of exceptional free cash flow while retaining enough capital to pursue the next phase of AI-memory demand.
SK Hynix’s board approved a 40 trillion won buyback, equivalent to about $28.6 billion. The company planned to repurchase up to approximately 24.07 million shares on the open market from August 20 through November 19, then cancel all of the shares acquired. The amount represented about 3.3% of shares outstanding.
The cancellation matters because it permanently reduces the company’s share count. If earnings remain unchanged, fewer shares divide the same earnings base, potentially increasing earnings per share. The buyback also returned cash directly to shareholders who sold into the program, while the cancellation provided a stronger per-share signal than holding repurchased shares in the treasury.
SK Hynix simultaneously raised its medium-term shareholder-return target. For cumulative free cash flow generated from 2025 through 2027, it moved from returning “within 50%” to returning more than 50% through repurchases, cancellations and dividends. The company also left open the possibility of a special dividend and further measures.
SK Hynix said its intrinsic value—based on its business competitiveness and cash-generation capability—was not fully reflected in its current share price. In practical terms, management was signaling that it viewed the post-selloff valuation as too low and was willing to use its balance sheet to act on that view.
The timing reflected a clash between record AI-related profits and investor concerns that the boom could be approaching a peak. Shareholders wanted a larger share of the cash being generated by the memory cycle, while chip stocks fell as markets questioned whether AI infrastructure spending and memory prices could remain elevated.
SK Hynix had fallen more than 50% over two months before announcing its buyback, according to reporting on the plan. Investors had also focused on the companies’ large expected cash balances and their existing policy of returning roughly half of free cash flow, arguing that more could be distributed after the AI-driven earnings surge.
The response therefore served two purposes:
But the measures did not prove that memory prices would continue rising. A buyback reduces the number of shares and signals management confidence; it does not determine future supply, pricing or demand. Those fundamentals still depend on AI capital expenditure, memory supply discipline and broader financial conditions.
Investors broadly treated SK Hynix’s cancellation as a strong vote of confidence in the company’s near-term business outlook and AI-server demand. The size and speed of the program suggested that management considered the market’s valuation concerns excessive relative to the company’s operating prospects.
The interpretation was more nuanced than “buybacks guarantee higher chip prices.” The announcement could support the stock through a smaller share count and improved confidence, but memory remains a cyclical industry. If AI spending slows or supply expands faster than demand, shareholder returns would not prevent pressure on profits.
Samsung’s plan carried a related message, although its larger emphasis on dividends made it a different type of response. Cash dividends provide a direct payout, while the company’s retention of investment capacity kept the focus on whether it could capture continuing demand for high-bandwidth memory and data-center chips.
The announcements were followed by a sharp rebound in South Korean equities. On August 20, SK Hynix rose 12.73% to 1.691 million won and Samsung Electronics gained 9.49% to 271,000 won. The Kospi advanced 5.9% and moved back above 6,800.
The rally reflected both the shareholder-return news and a broader improvement in risk appetite as U.S. Treasury yields eased. Higher long-term Treasury yields had previously weighed on growth and semiconductor valuations, so the market reaction was not solely a judgment on the companies’ payout policies.
The announcements also helped support a wider semiconductor rebound. Still, analysts continued to watch two unresolved risks: whether high U.S. yields would pressure technology valuations again and whether a future slowdown in AI infrastructure spending would weaken memory demand.
The scale of the plans was made more credible by the companies’ strong cash-generation backdrop. Reuters reported that Samsung and SK Hynix together were expected to hold $263 billion in net cash by year-end.
Micron was also discussed as a comparison after pledging to return 100% of “excess cash” to shareholders over the long run, according to reporting cited in the source material. That is not directly equivalent to Samsung’s 50%-of-free-cash-flow framework or SK Hynix’s more-than-50% cumulative target: “excess cash” and free cash flow are different capital-allocation concepts, and the available reporting does not establish identical definitions or timetables.
The August measures changed the debate around Korea’s memory-chip leaders from whether they would share their AI windfall to how much they could return while maintaining the investment required for the next memory cycle.
Samsung offered a large, dividend-led return with continued reinvestment. SK Hynix delivered a more immediate valuation signal through a 40 trillion won repurchase and cancellation, while lifting its cumulative payout target above 50% of free cash flow. Both approaches supported their stocks in the short term, but neither removed the fundamental test ahead: whether AI infrastructure demand and advanced-memory pricing can justify the earnings expectations embedded in the companies’ valuations.