#海力士回购落地
SK Hynix $SKHY suddenly announced a share buyback plan worth as much as 40 trillion KRW, and it is not just a simple repurchase to hold on the books, but a clear plan to repurchase and cancel the shares.
At the same time, the company’s 2026 capital expenditure plan has instead been raised further, with expectations reaching 40 trillion KRW. Looking at these two figures together is much more interesting than a simple “buyback positive” story. On one side, it is expanding aggressively; on the other, it is taking out a large amount of cash to reward shareholders. What does that mean? At the very least, it shows one thing: the memory-market boom brought by AI has already generated enough cash flow for SK Hynix, enough that even while continuing to increase investment in advanced capacity such as HBM, it still has room to share profits with shareholders. And after the buyback, the shares are canceled directly, which is not the same as an ordinary treasury share repurchase.
According to the currently disclosed plan, the shares to be canceled are expected to account for about 3.3% of total shares outstanding. If future net profit remains unchanged, then after the denominator is reduced, EPS will naturally be lifted in theory. But I think the thing worth paying the most attention to is not whether EPS rises by a few percentage points. The real question is: why is SK Hynix willing to take out so much money now? Because what the market is truly worried about has never been whether SK Hynix can make money this year, but how long the AI boom can last.
In the past, the memory industry had a very typical cycle: demand picks up → everyone expands aggressively → oversupply → prices collapse → profits return to square one (I have written about this logic many times).
This time, HBM has indeed brought unprecedented demand to the memory industry, but if cloud companies’ capital expenditures slow down in the future, or if the supply-demand gap for HBM gradually narrows, can today’s high profits still be sustained?
So SK Hynix’s large-scale buyback now is actually also giving the market a reassurance: the money it has earned will not all be turned back into ammunition for the next round of capacity warfare. Of course, a buyback itself is not a cure-all. If the industry really is at the top of the cycle, even if the company buys back all the shares, it still cannot stop product prices from falling and profits from declining. So what I will pay closer attention to next are three things:
1. whether HBM demand can keep growing;
2. how long DRAM prices can be maintained;
3. and whether the capital expenditures of AI giants have shown a clear inflection point.
$SKHY
SK Hynix $SKHY suddenly announced a share buyback plan worth as much as 40 trillion KRW, and it is not just a simple repurchase to hold on the books, but a clear plan to repurchase and cancel the shares.
At the same time, the company’s 2026 capital expenditure plan has instead been raised further, with expectations reaching 40 trillion KRW. Looking at these two figures together is much more interesting than a simple “buyback positive” story. On one side, it is expanding aggressively; on the other, it is taking out a large amount of cash to reward shareholders. What does that mean? At the very least, it shows one thing: the memory-market boom brought by AI has already generated enough cash flow for SK Hynix, enough that even while continuing to increase investment in advanced capacity such as HBM, it still has room to share profits with shareholders. And after the buyback, the shares are canceled directly, which is not the same as an ordinary treasury share repurchase.
According to the currently disclosed plan, the shares to be canceled are expected to account for about 3.3% of total shares outstanding. If future net profit remains unchanged, then after the denominator is reduced, EPS will naturally be lifted in theory. But I think the thing worth paying the most attention to is not whether EPS rises by a few percentage points. The real question is: why is SK Hynix willing to take out so much money now? Because what the market is truly worried about has never been whether SK Hynix can make money this year, but how long the AI boom can last.
In the past, the memory industry had a very typical cycle: demand picks up → everyone expands aggressively → oversupply → prices collapse → profits return to square one (I have written about this logic many times).
This time, HBM has indeed brought unprecedented demand to the memory industry, but if cloud companies’ capital expenditures slow down in the future, or if the supply-demand gap for HBM gradually narrows, can today’s high profits still be sustained?
So SK Hynix’s large-scale buyback now is actually also giving the market a reassurance: the money it has earned will not all be turned back into ammunition for the next round of capacity warfare. Of course, a buyback itself is not a cure-all. If the industry really is at the top of the cycle, even if the company buys back all the shares, it still cannot stop product prices from falling and profits from declining. So what I will pay closer attention to next are three things:
1. whether HBM demand can keep growing;
2. how long DRAM prices can be maintained;
3. and whether the capital expenditures of AI giants have shown a clear inflection point.
$SKHY