Larry Ellison was the richest person on the planet for one afternoon in September 2025. It’s not Elon Musk. Not Jeff Bezos. After Oracle announced a $300 billion computing contract with OpenAI, Ellison, an eighty-year-old co-founder of Oracle who spent decades developing database software that most people used without knowing his name, saw his net worth increase by about $100 billion in a single day. It was a fleeting moment. In a matter of hours, his position on the wealth list changed once more. However, the episode revealed something more enduring than any one man’s wealth: the AI boom has started to affect international financial markets in ways that have no clear historical precedents, and it is challenging to fully comprehend the scope of what is happening.
In 2025 alone, data center spending is expected to surpass $400 billion, which is roughly equivalent to Denmark’s entire GDP. By 2030, McKinsey projects that the total will be close to $7 trillion. According to Goldman Sachs Research, data center power consumption will increase by 165% from 2023 levels by 2030. These figures have that slightly surreal quality that very large numbers frequently have, where the numbers are technically understandable but the real economic and physical effects are more difficult to envision. In practical terms, this means that massive quantities of concrete, copper, cooling infrastructure, electricity generation capacity, and human labor are being diverted to the construction of the physical foundation upon which artificial intelligence operates. Unlike dot-com money, this money is not speculative. It is being used for things that actually exist.
Key Information: The AI Boom & Global Finance (2025–2026)
| Topic Focus | AI-driven restructuring of global financial markets, debt financing, and market concentration |
| CoreWeave Revenue vs. Spend | ~$5B revenue vs. ~$20B spending (2025) — carrying $14B in debt, $34B in scheduled lease payments by 2028 |
| CoreWeave Customer Concentration | Microsoft: ~70% of revenue; Nvidia + OpenAI: ~20% — extreme single-client dependency |
| Data Center Spend (2025 est.) | Conservatively projected to exceed $400 billion — roughly the size of Denmark’s entire economy |
| Data Center Spend (McKinsey est.) | Projected to reach ~$7 trillion cumulatively by 2030 |
| Global Power Demand Forecast | Data center power demand forecast to rise 165% by 2030 vs. 2023 levels (Goldman Sachs Research) |
| Magnificent Seven Market Share | ~39% of total S&P 500 value (2025 YTD); ~74% of NASDAQ 100 — unprecedented concentration |
| Oracle’s Single-Day Wealth Event | Share price rose 43% in one day after OpenAI deal worth $300B in computing — briefly made Larry Ellison world’s richest person |
| OpenAI–Oracle Deal | $300 billion in computing power over five years — one of the largest AI infrastructure contracts ever announced |
| Nvidia Market Position | Controls ~90% of the AI chip market; trades at 30x+ forward earnings; central node in circular AI financing arrangements |
| AI Companies Using AI in Finance | 71% of surveyed companies using AI across corporate finance (KPMG Global AI in Finance Report) |
| Historical Parallel | Dot-com crash of March 2000: NASDAQ fell 77% over two years; AI concentration mirrors late-1990s levels |
| Key Difference from Dot-Com Era | Today’s AI leaders (Microsoft, Apple, Google) are profitable with real revenue — not cash-burning startups |
| Key Risk Flagged | Circular financing arrangements between AI companies, lenders, and chip suppliers — echoing pre-2008 financial engineering complexity |
However, the financial structures that underlie all of this are worth closely examining because they have started to resemble something other than simple capital investments in some ways. The most obvious example is CoreWeave. CoreWeave, a former cryptocurrency mining company that switched to data center operations, raised money, purchased Nvidia chips in bulk, constructed facilities to house them, and rented the entire setup to AI companies that required processing power without having to pay for it up front. That business plan makes sense.
The financial architecture that surrounds it is less understandable. CoreWeave anticipated earning about $5 billion in revenue in 2025, spending about $20 billion, and owning $14 billion in debt, the majority of which was arranged through newly established legal entities made expressly to borrow money from private equity lenders at high interest rates on the company’s behalf.
Additionally, between now and 2028, $34 billion in scheduled lease payments are due. Microsoft accounts for up to 70% of the company’s revenue, making them its main client. Nvidia, a major investor and supplier of chips to CoreWeave, is among its next largest clients. This means that CoreWeave is essentially using Nvidia’s funds to purchase Nvidia’s chips and lease them back to Nvidia. In addition to being a significant investor and customer of CoreWeave, OpenAI has strong financial ties to Microsoft and Nvidia. The last time so much wealth was entangled in such layered, overlapping financial structures was shortly before the 2008 financial crisis, according to a December 2025 article in The Atlantic about these arrangements. Even though the comparison is not exactly comparable, it is still worthwhile to consider.
The question about concentration is just as awkward. The Magnificent Seven—Apple, Microsoft, Alphabet, Amazon, Meta, Tesla, and Nvidia—now make up about 74% of the NASDAQ 100 and roughly 39% of the S&P 500. This means that, whether they intend to or not, an investor purchasing a typical S&P 500 index fund—generally regarded as a diversified investment—is making a significant and increasing wager on artificial intelligence.
Although not perfect, there is a clear parallel to the late 1990s. Technology valuations at the time were driven to levels that had little to do with real business fundamentals due to dot-com euphoria. Trillions of dollars’ worth of paper wealth were destroyed and a significant amount of real wealth was lost when the Nasdaq finally corrected in March 2000, falling 77% over the course of two years. The key distinction between today’s AI leaders is that they are successful, well-established companies with substantial current revenue streams. Amazon, Google, and Microsoft are not spending money to keep a story going. In addition to making huge bets on AI infrastructure, they are making significant profits. This time, the real business value and the speculative excess are mixed together in proportions that are actually difficult to separate.
It’s difficult to watch all of this without experiencing a certain slow-moving tension; it’s more subdued and unsettling than the loud alarm of an obvious crash signal. Early in 2026, the IMF observed that the surge in AI investment is changing the economic structure while also highlighting how inadequately current measurement systems reflect the real situation. This rate and scale of intangible capital formation was not intended to be tracked by GDP statistics. Financial arrangements of this complexity were not intended for regulatory frameworks. Furthermore, it is still genuinely unclear whether current valuations represent realistic expectations for future AI profitability or if they represent the kind of optimism that always seems reasonable until it doesn’t. The boom is genuine. There is a real reshaping. It’s still unclear what will happen next.
