AI Capitalism Pushes the Global Economy Toward $1.5 Trillion in Additional Debt
$1.5 trillion in additional debt over the coming years: this is the price global companies are preparing to pay to finance artificial intelligence infrastructure. This projection from Morgan Stanley and JPMorgan marks the world’s entry into a new economic era where AI becomes an infrastructure sector comparable to energy or telecommunications.
This shift is transforming the very nature of contemporary capitalism. American and Chinese technology giants, followed by governments and traditional companies, are borrowing heavily to build the data centers, chips, and networks that will power tomorrow’s digital economy.
The Essentials
- $1.5 trillion in additional debt projected to finance global AI infrastructure according to Morgan Stanley and JPMorgan
- Meta, Microsoft, and Amazon each invest more than $60 billion annually in their AI capabilities
- Data center electricity consumption is expected to triple by 2030, requiring 150 equivalent new nuclear reactors
- The debt-to-capitalization ratio of tech companies is reaching record levels, comparable to 19th-century railroad investment cycles
Meta Borrows $60 Billion to Catch Up to OpenAI
Meta illustrates this race into debt. Mark Zuckerberg’s company plans to spend between $60 and $65 billion in 2025, a 50% increase from 2024. This budget primarily finances the purchase of Nvidia H100 chips and the construction of giant data centers.
Microsoft follows with $64 billion in planned investments for 2025, while Amazon Web Services will spend $75 billion over the same period. These amounts exceed the infrastructure budgets of many developed nations. By way of comparison, France dedicates 54 billion euros annually to its public investments.
This escalation responds to competitive urgency. OpenAI, valued at $157 billion, has demonstrated that an AI model that excels can reshape the technological landscape in just a few years. Established companies are therefore accepting new levels of debt rather than risking obsolescence.
AI Already Consumes as Much as a Small Country
AI infrastructure is transforming global energy needs. A data center dedicated to training models like GPT-4 consumes 50 megawatts continuously, equivalent to 40,000 American households. Projections indicate this consumption will triple by 2030.
This demand is driving technology companies toward direct energy investments. Microsoft has signed a 20-year contract to restart the Three Mile Island nuclear plant, closed since 2019. Amazon is investing $500 million in new modular reactors. Google is financing four nuclear reactors to power its California data centers.
The International Energy Agency estimates that data centers will represent 6% of global electricity consumption by 2030, compared to 2% today. This progression is equivalent to adding the energy demand equivalent of Japan to the global electricity grid.
Europe is struggling to keep pace with this rate of energy investment, as shown by its strategy for economic repositioning in the face of contemporary challenges. European regulatory constraints are slowing the approval of new reactors, creating a competitive disadvantage in the race for AI.
China Doubles Its Computing Capacity Without American Chips
China is financing its technological independence through coordinated state debt. Beijing is injecting $143 billion into its semiconductor sector in 2025, primarily through state-guaranteed bank loans.
This strategy is producing tangible results. The construction of the Tianhe-3 supercomputer with 2 exaflops without American components demonstrates the effectiveness of this centralized financial approach.
Chinese companies are adopting hybrid financing models. Baidu is borrowing $12 billion to develop its data centers, with indirect support from the People’s Bank of China. Alibaba Cloud is raising $15 billion through convertible bonds to finance its expansion in generative AI.
This Sino-American race is creating financial fragmentation. Western banks are hesitant to finance projects involving Chinese technologies, while Chinese financial institutions refuse to invest in infrastructure dependent on American components subject to sanctions.
Central Banks Face a New Investment Cycle
Central banks worldwide are scrutinizing this explosion of private debt. The Federal Reserve estimates that AI investments now represent 23% of new corporate bond issuances, compared to 8% in 2022.
This concentration worries financial regulators. If a technology recession were to occur, the domino effect would touch the entire Western banking system. American banks hold $340 billion in claims related to technology infrastructure, or 15% of their corporate credit portfolios.
The European Central Bank is adopting a more cautious approach. Its stress tests now include scenarios for a sharp drop in technology valuations. European banks must provision 12% of their tech loans, compared to 8% in the United States.
This regulatory divergence is accelerating Europe’s decoupling. European companies have difficulty accessing the financing necessary for their AI projects, unlike their American and Chinese competitors who benefit from more flexible conditions.
The Emergence of a Digital Infrastructure Capitalism
This wave of debt marks the emergence of a new economic model. AI becomes a critical infrastructure sector, comparable to 19th-century railroads or 20th-century highways. Companies that control these infrastructures will hold a lasting competitive advantage.
Valuations reflect this structural transformation. Nvidia, manufacturer of chips essential to AI, displays a price-to-earnings ratio of 65, similar to railroad companies during their expansion. Microsoft and Google are restructuring their balance sheets to resemble utilities rather than traditional software companies.
This evolution redefines global financial risks. The interdependence of AI infrastructures creates new systemic vulnerabilities. A major failure at a chip supplier or data center can paralyze entire sectors of the digital economy.
Governments are beginning to treat AI as a public service. The Biden administration classifies data centers as critical infrastructure, at the same level as power plants. China is integrating its computing capacity into its national strategic planning.
This mutation of capitalism toward an infrastructure model financed by debt is accelerating. The $1.5 trillion in planned debt represents only the beginning of an investment cycle that could last a decade. The companies and nations that master this financing will hold the keys to tomorrow’s economy.