The Unprecedented Flow of Capital into AI
The artificial intelligence industry is currently experiencing a financial boom unlike any other, with massive amounts of capital pouring into the infrastructure required to develop and power these new technologies. What began as an internal funding effort by tech giants has rapidly expanded to include mainstream credit investors, private credit lenders, and bond investors.
For example, JPMorgan Chase and Mitsubishi UFJ Financial Group are leading a colossal $22 billion loan to back Vantage Data Centers’ construction of a new campus, while Meta Platforms is securing a $29 billion investment from major players like Pacific Investment Management Co. and Blue Owl Capital Inc. for its data center in Louisiana. This influx of capital is driven by the industry’s estimated need for trillions of dollars to support future AI services, signaling a belief that this is the next major economic frontier.
Echoes of the Dot-Com Bubble
Despite the overwhelming optimism, many seasoned industry experts and analysts are drawing uneasy parallels between the current AI investment frenzy and the dot-com bubble of the late 1990s. OpenAI CEO Sam Altman himself has acknowledged the resemblance, cautioning that “someone’s gonna get burned.” The primary concern is that a great deal of money is being invested in a technology whose long-term revenue potential is still largely unproven.
A report from a Massachusetts Institute of Technology initiative, for instance, indicated that a staggering 95 percent of generative AI projects in the corporate world have yet to yield a profit. This historical context from the early 2000s, when overbuilt telecom infrastructure led to significant write-downs, serves as a cautionary tale for credit investors who are now financing the massive build-out of AI infrastructure.
Private Credit and the New Funding Model
The shift in funding sources is a crucial aspect of this investment boom. While the initial build-out was funded by tech companies’ internal capital, the burden is now increasingly falling on private credit lenders and public bond markets. Private credit has become an especially significant player, with funding for AI running at an estimated $50 billion per quarter over the last three quarters alone. This far surpasses the capital being provided by public markets, even without accounting for the mega-deals from Meta and Vantage.
The exposure for these investors comes with varying degrees of risk. While debt from “AI hyperscalers” like Google and Meta is generally considered safe due to their existing cash flows, much of the new money is coming from private credit markets, which operate with less transparency and regulation.
The Role of Commercial Mortgage-Backed Securities
Adding another layer of complexity to the financial landscape is the increasing use of commercial mortgage-backed securities (CMBS) to fund new computing hubs. Instead of being tied to a corporate entity, these securities are backed by the payments generated by the data centers themselves. According to a recent JPMorgan Chase & Co. estimate, the amount of CMBS backed by AI infrastructure has already surged by 30 percent this year, reaching $15.6 billion.
This funding model transfers the risk from corporate balance sheets to the performance of the data centers, creating a new set of concerns for analysts. The long-term viability of these assets, which are being funded for a 20- to 30-year tenor, is a major unknown, as the technology itself is expected to change dramatically in the next five years.
Assessing Long-Term Revenue and Sustainability
A central concern among financial analysts is the sustainability of the current investment model. As S&P Global Ratings’ Ruth Yang pointed out, there is no historical basis for assessing the forward cash flows of these new AI ventures. The risk is that these deals are being financed before AI projects have demonstrated a long-term ability to generate revenue.
This is a particularly pressing issue for utility firms that are borrowing heavily to build the electrical infrastructure required to power these energy-hungry data centers. The stress on the system is already starting to show, with a rise in payment-in-kind (PIK) loans, where interest payments are made with more debt rather than cash, a trend that often signals financial strain.
The Evolving Nature of AI Infrastructure
The physical and technological demands of AI are changing at a dizzying pace. The infrastructure being built today may quickly become obsolete as new models and hardware emerge. This presents a significant challenge for long-term financing. While direct lenders see the massive capital need of hyperscalers as a new long-term infrastructure asset, the reality is that the technology is far from static.
Unlike a stable asset like a highway or a bridge, a data center built to run a specific type of AI model might not be able to accommodate the next generation. This uncertainty makes the assessment of long-term risk and return incredibly difficult for even the most experienced credit watchers.
Navigating the Future of AI Investment
The investment boom into artificial intelligence is a clear signal of its transformative potential, but it is not without significant risk. The shift from corporate funding to private credit and CMBS introduces new vulnerabilities to the financial system. While some of the debt may be secured by gold-plated corporate entities, a large portion is tied to the uncertain future cash flows of a rapidly evolving technology.
As the fire hose of money continues to pour into the sector, driven by a search for the next long-term infrastructure asset, investors and analysts will need to carefully navigate the parallels to past bubbles and scrutinize the true long-term sustainability of these projects. The future of AI investment will depend on a careful balance between seizing opportunity and exercising caution.
Read More: AI in Science: Useful or Overhyped? The Debate Continues
