The Next USD “Peg”: Why the AI Arms Race is a Defense Budget
The “Unipolar Moment” is not over; it has just moved up the technology stack. The U.S. currently maintains its hegemony through three distinct “Pegs”—monetary, digital, and intelligence. The U.S. is currently spending trillions on AI CapEx not because it is profitable in the short term, but because it is the only way to defend the third and most critical peg.
1. The Monetary Peg (The Dollar)
- The Metric: Transaction Volume.
- The Data: As of January 2025, the U.S. Dollar accounted for 50.2% of all global payment traffic routed via SWIFT.
- The Reality: Despite talk of “de-dollarization,” the USD remains the operating system for half of global trade.
2. The Digital Peg (The Cloud)
- The Metric: Cloud Infrastructure Market Share (The servers where the internet “lives”).
- The Data: As of Q1 2025, just three U.S. companies (Amazon AWS, Microsoft Azure, Google Cloud) control approximately 66% of the global cloud infrastructure market.
- The Reality: While “internet traffic” paths are decentralizing, the destination of that traffic is centralizing. Two-thirds of the world’s digital economy runs on computers owned by three American corporations.
3. The Intelligence Peg (AI Models)
- The Metric: Global Generative AI Traffic Share (The “brains” processing the world’s queries).
- The Data: As of late 2025, U.S.-owned models (OpenAI/ChatGPT, Gemini, Copilot, Claude) account for approximately 89% of global generative AI web traffic and API volume.
- The Reality: The concentration of power increases as you move up the stack. While the world uses the Dollar for 50% of trade and U.S. clouds for 66% of storage, it relies on U.S. algorithms for nearly 90% of its synthetic intelligence.
The Explainer: Why Trillions in CapEx?
The U.S. is projecting over $1.4 trillion in AI CapEx through 2035. This is effectively a defense budget designed to counter two specific threats:
1. Countering “Efficient” Sovereignty: Chinese models (like DeepSeek) have proven AI can be built cheaply. If costs stay low, nations will build their own “Sovereign AI” and disconnect from the U.S. stack. The U.S. must spend to keep its models so advanced that sovereign alternatives remain “second-tier.”
2. CapEx as a Moat: By pushing the price of frontier intelligence to hundreds of billions, the U.S. creates a “Compute Moat.” This prices out other nations, forcing them to rent intelligence from American APIs rather than building their own.
The debt is the cost of hegemony. The U.S. is socializing the risk to ensure that in 2030, the world doesn’t just transact in Dollars, but thinks in American.
