12:25 pm - April 16, 2026For several months now, a picture has been subtly circulating in tech and financial circles. From above, it depicts an Amazon Web Services data center in Manassas, Virginia, with rows of identical gray buildings perched on level ground, resembling a tiny city that exists solely for thought. On the surface, the image is unremarkable.
The narrow sea passage in the Persian Gulf, the gas terminals in Qatar, the chip factories in Taiwan, and the ships crawling through the Strait of Hormuz are all examples of what makes it striking. It’s all invisible. Everything is crucial. Everything is in some sort of danger at the moment.
| Topic | Details |
|---|---|
| Subject | AI Boom Vulnerability & Multidimensional Economic Crisis (2026) |
| Core Thesis | Global economy dangerously concentrated in AI investments; supply chain exposed to Middle East instability |
| Key Authors | Matteo Wong & Charlie Warzel, The Atlantic (March 26, 2026) |
| AI Investment Scale | Trillions of dollars globally; in late 2025, virtually all U.S. economic growth came from AI investment |
| Critical Supply Chain Nodes | Advanced chips: primarily two South Korean companies + one Taiwanese firm |
| Energy Dependency | South Korea & Taiwan source majority of crude oil and LNG from Persian Gulf |
| Key Chokepoint | Strait of Hormuz — described as critical to basically every aspect of the global economy |
| Oil Price Trajectory | Brent crude climbed from $61.41 on January 1, 2026, to $112.57 as of March 30, 2026 |
| Financial Risk | AI firms carrying historic debt loads; banks and private credit deeply interlinked |
| Token Economics Risk | Cost per AI token falling rapidly — described by analysts as a death spiral to zero |
| Historical Parallel | Compared structurally to 2008 financial crisis dynamics |
| Key Voices | Sam Winter-Levy (Carnegie Endowment); Brad Lipton (Roosevelt Institute) |
Investments in artificial intelligence accounted for practically all of the US economy’s growth in the last months of 2025. Not most growth. A plurality, not. Almost everything. That statement merits a moment of silence because it depicts something that would have seemed unthinkable even five years ago: a single industry subtly taking over as the entire engine of the economy, heavily financed by debt, and based on a supply chain that crosses some of the world’s most unstable geopolitical regions.
By late March, Brent crude had risen from $61.41 at the beginning of 2026 to $112.57. The ascent has been relentless rather than gradual, motivated more by the expectation of supply cuts than by actual ones. Fear has economic weight because markets now price geopolitical risk in real time.
Traders don’t hold off until the Strait of Hormuz closes. They factor in the possibility that it could. And those odds are not insignificant in the Middle East right now.
Here, the dependencies were never truly concealed. Since at least the early 2020s, policy analysts have written about the geographic concentration of chip manufacturing. Taiwan Semiconductor Manufacturing Company’s unique position in the global supply chain was impossible to overlook due to the COVID-era shortages.
It was also widely known that South Korea and Taiwan, the two countries most in charge of manufacturing the sophisticated chips needed for AI model training, import most of their liquefied natural gas and crude oil from the Persian Gulf. There was always the math. Perhaps nobody wanted to complete the equation.
The Strait of Hormuz is essential to almost every facet of the world economy, according to Sam Winter-Levy, a technology and national security researcher at the Carnegie Endowment for International Peace. Additionally, he has pointed out that the AI supply chain was never safeguarded, with the same composure that professionals use when making assessments that are truly concerning.
The business just carried on as if it were. Businesses issued debt. Flat land in rural Texas, Virginia, and Nevada gave rise to data centers. The build-out proceeded at a rate that presumed constant, unbroken access to chips, energy, and the shipping routes that connected them.
This unfolds in a certain way and follows a structural logic. In manufacturing economies in East Asia, energy reserves are measured in months rather than years. Semiconductor production starts to slow if the Strait’s traffic is seriously disrupted long enough for those reserves to run out. The already tight timelines for building data centers will get longer if semiconductor production slows.
The AI companies that raised capital under the assumption of exponential, uninterrupted growth will suddenly face a very different set of calculations than those outlined in their pitch decks if data center timelines slip.
Brad Lipton, a former senior adviser at the Consumer Financial Protection Bureau who is currently employed at the Roosevelt Institute, has identified structural similarities between the current situation and the pre-2008 period: banks lending to private credit funds, which lend further down the chain, with the underlying risk increasing rather than decreasing at each stage.
Large sums of money were drawn to the AI industry by investors who anticipated profits from businesses that aren’t yet profitable in most conventional accounting terms.
The price of each AI-generated token, which is the essential component of what these companies sell, has been falling quickly in the meantime. For the revenue models that support the entire investment thesis, some analysts characterize this as a death spiral toward zero. Whether efficiency improvements will eventually create enough new demand to save the economy or if the math was always a little too optimistic is still up for debate.
As this develops, it’s difficult to ignore the 1970s echo beneath it. The Iranian Revolution of 1979 and the OPEC embargo of 1973 both demonstrated how quickly energy disruption spreads throughout an economy, causing growth to slow, inflation to rise, and central banks to be caught between worse options. Now, speed is the difference.
What used to take months to filter through economies can now happen in a matter of weeks. Both executives and policymakers seem to be functioning within a framework designed for a slower world, one in which there was time for gradual repositioning, hedging, and adjustment. That margin has shrunk considerably.
Technically, there is a version of this that doesn’t end badly. The financial exposure spreads throughout a system big enough to absorb the losses without cascading, the Strait reopens swiftly, chip supplies stabilize, and AI companies find revenue models that justify their debt loads.
Many things would need to go smoothly, in the correct sequence, and with minimal additional disruption. It is feasible. However, as Warren Buffett once said, you only find out who has been swimming unprotected when the tide goes out. The beach is packed right now, and the water is clearly receding.
