Bloomberg: AI Boom Has Pushed Global Trade Imbalances to Their Widest Since 2008

Global trade imbalances have reached levels not seen since the 2008 financial crisis. In a new analysis, Bloomberg Originals names the artificial intelligence boom as the main cause: it has gathered enormous wealth in one place and shifted global flows of goods and capital. The announcement itself is short and discloses no specific figures, so below we walk through the mechanics behind the claim and show which indicators can help test it.

How AI Skews Trade

AI infrastructure is physical. Every new cluster means accelerators, memory, server racks, networking equipment, transformers and cooling. Most of this is made in a limited number of places: chips and packaging in Taiwan, memory in South Korea, server assembly in Taiwan, China, Mexico and Vietnam. The buyers, meanwhile, sit mainly in the US, where the largest cloud providers spend hundreds of billions of dollars a year on data centers.

As a result, equipment imports into the buyer's country rise, and exporters build up surpluses. This is the classic mechanism of a trade imbalance. It used to be fed by cars and consumer electronics; now it is fed by racks of GPUs.

Where the Money Ends Up

The second part of the story concerns capital. AI profits are distributed unevenly: they go to accelerator makers, cloud platforms and a handful of labs that train frontier models. Nvidia's market capitalization topped $5 trillion for the first time in October 2025, and that figure describes one company, not the whole market. The investment is funded partly by outside capital, so money flowing into US assets feeds the same asymmetry from the other side of the balance of payments.

The result is a closed loop. Capital flows into US AI assets, the US buys equipment in Asia, Asian exporters accumulate foreign-currency revenue, and some of it goes back into dollar securities. As long as profit expectations are met, the loop holds. If they are not, the imbalance will be financial as well as trade-related.

What to Watch in the Data

• The threshold: by Bloomberg Originals' estimate, imbalances are at their highest since 2008, that is, in 18 years.
• Hyperscaler capex: quarterly reports from Microsoft, Alphabet, Amazon and Meta show how much money is going into hardware.
• Electronics exports: statistics from Taiwan, South Korea and Vietnam on servers, chips and memory.
• Concentration: the share of a few companies in index market capitalization, above all the S&P 500.
• Funding sources: debt issuance for data centers and private credit.

Practical Takeaways

For companies building products on AI, the main risk sits in the supply chain. The price and availability of accelerators depend on a few fabs and on trade policy. It makes sense to lock in cloud capacity contracts for the long term and keep a plan ready in case compute prices are repriced.

Investors will want to separate revenue from real demand for AI services from revenue that rests on cyclical capex. The former shows the industry's staying power; the latter depends on the mood of the capital markets.

Finally, trade imbalances of this size tend to provoke a political response sooner or later: tariffs, export restrictions, subsidies for local manufacturing. For the AI industry, this is no longer a hypothesis but a working planning scenario. The full Bloomberg analysis with figures is worth reading in the original, to check our caveats against the data in the piece itself.