Heavy borrowing to fund artificial intelligence (AI) infrastructure is raising concerns about a potential market bubble. CoreWeave shares plunged 15% in one day as bond yields approached 15%, signaling greater perceived default risk.
Oracle shares also fell sharply. Investors are selling AI-related debt, demanding higher returns and questioning whether future profits will justify massive spending.
Rising interest rates, inflation and increased bond supply are adding pressure. CoreWeave faces significant financial risks. It buys expensive Nvidia chips, leases data centers long-term and rents computing capacity to customers such as OpenAI, Microsoft and Meta. This leaves it responsible for major costs even if customers reduce demand or leave before the leases expire. Its operating margins are around 4%, while quarterly interest expense reached $640 million, more than double the prior year. This makes CoreWeave’s financials look more and more like a WeWork for data centers.
Construction and permitting delays have also slowed projects. CoreWeave says its $104 billion backlog supports its debt and that customer contracts help repay financing, but investors remain concerned about profitability and cash flow.
Oracle faces similar challenges. Its cloud backlog is substantial, but fulfilling contracts requires enormous investment. The company has reported negative free cash flow, rising interest costs and heavy capital spending. Oracle borrowed $18 billion in September and may need significantly more debt to expand its AI capacity.
Project delays, power-permitting problems and reliance on major customers such as OpenAI add uncertainty. Oracle says it intends to maintain its investment-grade credit rating.
The risks extend across the industry.
Major technology companies issued more than $228 billion in bonds by mid-October, while AI infrastructure investment could exceed $1 trillion annually by 2027. SoftBank has also borrowed heavily to invest in OpenAI, which faces substantial cash needs and a reported $207 billion funding gap through 2030.
If AI spending slows down or funding dries up, data-center operators, chipmakers, lenders and investors could suffer losses. Three outcomes remain possible: AI profits justify current spending, valuations decline gradually, or funding problems trigger a broader selloff.
Larger companies such as Microsoft, Alphabet, Amazon and Meta have stronger finances than many smaller providers, and AI demand may continue growing. However, demand alone does not guarantee profits.
Investors should monitor the situation to determine if debt, borrowing costs, customer concentration, construction delays and free cash flow can continue to support AI growth or if the industry's mounting financial commitments may become unsustainable in the future.