As investment risks in AI infrastructure mount, Wall Street has begun incorporating GPU computing power rental prices into financial derivative trading. Trading institutions are developing products such as options linked to computing power prices to help data center operators hedge against declining rental income. At the same time, trading in technology stock put options and credit default swaps (CDS) is becoming increasingly active, reflecting growing market concerns over massive AI capital expenditures and investment returns.
GPU rental rates emerge as a new underlying asset for financial derivatives
Quantitative trading firm Jump Trading recently established a dedicated computing power trading division, exploiting price differences for computing power across different service providers to conduct trades and designing over-the-counter derivatives to provide price protection for AI infrastructure operators. Specifically, a cloud computing company that rents out Nvidia GPUs can purchase put options linked to computing power rental prices. If computing power supply increases in the future and rental prices fall sharply, the company can receive compensation as stipulated in the contract, thereby reducing revenue losses. The counterparty can either purchase computing power at an agreed price or settle the transaction in cash. Currently, the computing power rental price for Nvidia's B300 chip has risen about 56% this year to $7.56 per hour. One potential trade arrangement proposed by Jump Trading is to provide operators with price protection at $4 per hour, enabling them to hedge losses when rental rates fall below the agreed level. Another major trading institution, Susquehanna International Group, also said it has received inquiries from data centers and private AI cloud computing companies about hedging chip and computing power price risks. The related products may be linked to computing power price indices to provide protection for future rental income. These financial instruments could also improve financing conditions for AI infrastructure companies. For operators that rely on borrowing to purchase GPUs and build data centers, computing power rental income is an important source for repaying debt. If they can lock in part of their future income through derivatives, companies may have the opportunity to reduce lenders' concerns about cash flow volatility and thereby obtain more favorable financing terms. This trend also aligns with Nvidia's direction of promoting the financialization of computing power assets. In August this year, Nvidia CEO Jensen Huang proposed treating AI computing power as an infrastructure asset that can continuously generate income and be reallocated among different customers, and promoting long-term financing support from financial institutions for AI infrastructure. Today, derivatives developed around computing power rental prices are further bringing this asset class into financial trading markets.
AI investment risks rise, with hedging demand growing in both equity and credit markets
The rise of computing power derivatives comes as Wall Street reassesses AI investment risks. As technology companies continue to invest enormous sums in building data centers and purchasing AI chips, investors are increasingly focused on whether these investments can generate sufficient revenue and whether the companies involved can bear growing debt. Recently, AI-related company stock prices have pulled back sharply, driving up demand for put option trading. Data shows that put option trading volume for U.S. AI cloud computing provider CoreWeave hit a record high in October. Popular contracts included put options expiring in 2028 with a strike price of $35, while the company's recent stock price was around $55. The credit market has also shown signs of increased hedging demand. Credit default swap (CDS) spreads for Amazon, Alphabet, Microsoft, Meta and Oracle have generally widened over the past month, rising to their highest levels since 2021. This means investors must pay higher costs to hedge against the debt default risk of the companies involved. Among them, Oracle's CDS spread widened by more than 100 basis points, a particularly notable change. On October 30, the Nasdaq Composite Index fell about 2.9%, bringing its cumulative decline for October to about 5.3%, the worst monthly performance since March 2025. The continued correction in technology stocks further highlights investors' concerns about AI investment returns and financing sustainability. From stock prices and corporate debt to GPU rental income, Wall Street is expanding the scope of hedging against AI investment risks. However, the computing power derivatives market is still in an early stage of development, with trading scale far smaller than the overall AI investment market, and its liquidity and pricing mechanisms have yet to mature.
Comments