Samsung Electronics ended September holding 214.8 trillion won in cash and short-term financial assets, the company disclosed this week. That is about $146 billion, an increase of 24.8 trillion won in three months, and close to double the figure a year earlier. No Korean company had ever crossed 200 trillion won before.
The balance sheet around the pile is unusually clean. Net cash stood at 197.6 trillion won against total assets of 791.8 trillion, and the debt ratio fell from 31% to 29%. Samsung is not borrowing to fund its way through a downturn, and it is not holding cash because lenders have closed their doors. It is holding cash because the memory business, in the middle of an upcycle driven by artificial intelligence, has been generating it faster than the company has found ways to spend it. That sounds like a problem a chief financial officer would want. In the semiconductor industry it is a decision that has been deferred, and the deferral now has a size.
Three uses compete for the money, and Samsung has committed to two of them without choosing between the third. The first is shareholder returns. In August the company announced a programme worth 90 trillion to 110 trillion won, and management has said the cash balance is sufficient to cover it. At the low end that is more than 40% of the pile, and it is money that leaves the company permanently. Buybacks and dividends buy patience from investors who have spent years arguing that Korean equities trade below the value of their assets, a discount Samsung has done little to close.
The second is capital expenditure. Reports from Korea this week put the scale of EUV lithography orders placed by Samsung and by SK Hynix at 10 trillion to 20 trillion won each, with the larger figure at the top of the range. Those machines have lead times measured in quarters and are paid for over years. The cash on the balance sheet at the end of September is the same cash that will be spent on them, which means a large part of the pile is already spoken for and simply has not been invoiced.
The third is acquisitions. Samsung has a long record of preferring partnerships and internal development to large cheques, and buying at scale would be a departure from habit rather than a continuation of it. A balance sheet this large is the argument for trying anyway, because the memory cycle keeps showing that positions in high-bandwidth memory, advanced packaging and foundry capacity can be bought as well as built. Buying is faster and usually more expensive. The composition of the cash matters as much as the total. A won figure converts dollar holdings at the exchange rate on the reporting date, and the won has been weak against the dollar, which flatters the number in local currency. A company earning a market rate on 214 trillion won is generating interest income that flatters operating profit in quarters when chip prices are soft.
The strategic question is where Samsung wants to be in five years. Its foundry business trails the market leader by a margin that has widened rather than narrowed, and each missed node costs a customer relationship that is difficult to win back. Its memory business is profitable but cyclical, and the current cycle is being driven by demand from a small number of hyperscalers whose own capital spending is now a political topic. Its consumer businesses, phones and displays, generate revenue and thin margins. Against that, a cash pile is optionality. It lets Samsung keep investing through a downturn when competitors must retrench, which is how the company won previous cycles. It lets the company absorb a bad quarter without a capital raise.
The risk is that optionality becomes a habit. A balance sheet that large invites internal capital allocation by committee, where every division can argue for its share and none is forced to justify a return. Samsung’s disclosure that the shareholder return programme is covered by existing cash is a signal that management does not intend to raise money, but it is not a signal about where the rest goes. Nor does it say anything about the timing, and a return programme spread over three years is a different commitment from one executed in twelve months.
Investors will price the decision rather than the balance. A company that announces a 110 trillion won return and a 20 trillion won lithography order in the same season is telling the market that it has chosen both, at the cost of the acquisition option. A company that announces the return and stays quiet about the rest is telling the market it has not chosen at all. The first answer is expensive and defensible. The second is neither, and a number this size makes the silence costly.
