Credit investors are committing billions to AI infrastructure, even as concerns about a possible investment bubble grow.

JPMorgan Chase and Mitsubishi UFJ Financial Group are arranging over US$22 billion in loans for Vantage Data Centers to build a large data center campus, according to sources.

Meta is also securing US$29 billion from Pacific Investment Management and Blue Owl Capital for a major data center in Louisiana.

OpenAI has estimated it will need trillions in infrastructure investment to support its AI services.

Industry leaders, including OpenAI CEO Sam Altman, have compared the current frenzy to the dot-com bubble, warning that some investors could face losses.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ Massive infrastructure investment precedes proven revenue models, echoing historical overbuilding patterns

The current AI boom shows parallels to the dot-com era’s infrastructure overbuilding, with credit markets funding massive data center projects despite uncertain long-term returns.

JPMorgan and Mitsubishi are leading a $22 billion loan for Vantage Data Centers, while Meta secured $29 billion from PIMCO and Blue Owl for data centers in rural Louisiana1. These represent 20-30 year funding commitments for technology that may look entirely different in just five years, according to S&P Global Ratings1.

This mirrors the late 1990s telecom overbuilding, when business investment grew 10% annually fueled by technology optimism, leading to significant overinvestment and eventual writedowns2. The dot-com bubble saw the Nasdaq peak at 5,048 in March 2000 before declining 77% by October 2002, with much of the pain concentrated in infrastructure-heavy sectors3.

Citigroup’s credit strategists warn that “telecom companies arguably overbuilt and over borrowed and we saw some significant writedowns on those assets,” raising sustainability questions about current AI infrastructure spending1.

2️⃣ Credit markets continue massive AI funding despite widespread business execution failures

A disconnect exists between AI’s funding momentum and actual business results, with private credit flowing at $50 billion quarterly despite 95% of generative AI projects failing to generate profits1.

The MIT study revealing this 95% failure rate shows that companies have invested $35-40 billion in AI over the past year with limited measurable returns4. Most failures stem from misapplying AI to areas requiring human input, like sales and marketing, rather than focusing on back-office automation where AI excels4.

Yet private credit funding continues, representing “two to three times what the public markets are providing” according to UBS credit strategy1. This pattern of sustained investment despite poor fundamentals echoes the dot-com era, when venture funding continued flowing to unprofitable internet startups until the bubble’s eventual collapse.