NVIDIA, the world’s top AI chipmaker, which earns more than $100 billion a year (about 150 trillion won), is set to issue $20 billion (about 30 trillion won) in corporate bonds. It is a sign that even cash-rich companies are borrowing to compete for the AI market.
According to Reuters on the 15th (local time), NVIDIA will raise $20 billion in the U.S. bond market. Bonds are essentially debt certificates issued by a company to borrow money from investors. This offering will be divided into seven tranches with different maturities and terms, with the longest maturing in 2056.
This is NVIDIA’s return to the investment-grade bond market for the first time in five years. It marks its first borrowing since taking out $5 billion in June 2021.
The company said the proceeds will be used for general operating funds and for repaying or refinancing existing debt. Goldman Sachs, JPMorgan, and Morgan Stanley are managing the offering. NVIDIA shares rose 2% in early trading after the news broke.
◆ Even a company with abundant profits is issuing bonds: the logic of speed and scale
There is a reason why a company that earns so much would still choose to borrow. The answer lies in speed and scale.
The competition in AI semiconductors is a race in which companies must release stronger chips every year to survive. NVIDIA rolls out new chip lines annually, and each time it must front-load enormous development and production costs. No matter how large the profits are, that money is not immediately available in cash. As of the end of April, NVIDIA held $13.2 billion (about 20 trillion won) in cash, less than the $20 billion it is borrowing now.
Borrowing can also be the smarter choice. If interest rates are reasonable, a company can preserve its cash reserves and raise funds through bonds, leaving room for further investment. Interest expenses can also reduce tax burdens, and bonds can be used to refinance existing debt with better terms when it matures. For a cash-rich firm, issuing bonds is a strategy, not a sign of distress.
◆ AI investment powered by debt: Big Tech has all jumped in
NVIDIA is not alone. The scale of money being poured into AI is on another level entirely. Big Tech companies are expected to spend more than $700 billion (about 1,000 trillion won) on AI this year, nearly double last year’s $400 billion (about 600 trillion won).
The trend of raising firepower through bonds is also becoming clear. Meta launched a record bond sale of up to $30 billion in October last year, while Alphabet said last month that it would issue yen-denominated bonds for the first time. The astronomical cost of building data centers and buying chips has become difficult to cover with operating cash alone.
NVIDIA does not build data centers itself. But it makes the chips that go into them. As demand to train and run smarter AI has exploded, NVIDIA’s chips have become impossible to keep in stock. The fact that AI investment is increasingly relying on debt shows how hot the boom is, but it also leaves one question: will the returns really catch up with the money borrowed?
◆ A double-edged sword for Korea: HBM boom and bubble concerns
This trend is directly tied to Korea. NVIDIA’s AI chips use high-bandwidth memory, or HBM, and the companies that make it are SK Hynix and Samsung Electronics. The more chips NVIDIA produces, the fuller the coffers of Korean memory makers become. Big Tech’s $700 billion in AI spending is a direct tailwind for Korea’s semiconductor industry.
However, the same trend also brings risk. Because AI investment is increasingly being financed by debt, if expected profits fail to materialize and investment cools, the shock could spread to Korean companies supplying chips as well. It is a structure that carries both the gains of a boom and the danger of a bubble.
Ultimately, the key question is when “the money AI earns” will catch up with “the money poured in.” NVIDIA’s $20 billion bond issue is just one piece of that enormous bet. In a race where even the most profitable company is borrowing to keep up, the outcome will be decided not by investment, but by returns.