How Money Works asks what keeps America's AI boom financed
The essay's answer runs through chip-backed loans, hardware accounting and a late distinction between selling components and selling future possibilities.
Why this story? 2,155,944 views so far, and here is what is inside it.
Why China's AI Bubble Popped... And Ours Hasn't...
Next story: Krishnamoorthi compares advanced chips to aircraft carriers
Things we can learn from the video
Press to jump
- 3:40Chip purchases become assets whose costs are spread over time, rather than charged entirely against immediate profits.
- 5:40Longer depreciation schedules reduce annual accounting costs without reducing the cash already spent on hardware.
- 10:17CoreWeave illustrates how cloud companies use GPUs as collateral for borrowing.
- 20:00The narrator explains credit default swaps and how lenders seek protection against AI borrowers failing to repay.
- 23:39The title's answer contrasts Asian hardware suppliers' price exposure with American giants' ability to sell expectations of future AI.
A graphics chip can power an AI model. In How Money Works' essay, it can also secure a loan, making its future value a question for lenders as well as engineers.
"Why China's AI Bubble Popped... And Ours Hasn't..." has drawn more than two million views. Its central argument is that America's AI buildout is increasingly supported by borrowing and accounting assumptions, not simply the cash that technology companies already have.
The narrator explains how buying chips and data centers creates assets whose costs are spread across several years. Extending that accounting life makes annual profits look healthier, but it doesn't undo the money spent. The risk becomes sharper when companies such as CoreWeave borrow against their GPUs: falling hardware values could weaken the security behind those loans.
There is a complication, though. The essay describes older-chip rental prices recovering as computing supply tightens, helping sustain valuations even as newer hardware arrives. It then follows the financing outward to credit default swaps, contracts used to protect against borrowers failing to repay. The host argues that debt makes a gentle unwinding of AI investment harder to achieve.
The title's international comparison arrives around minute 24. The narrator's distinction is between component suppliers, exposed directly to prices and competition, and American technology giants that can keep investors focused on promises of future superintelligence. That is the essay's answer to why these markets need not deflate together, rather than a prediction of a particular crash date.
The video includes a paid sponsorship segment for a data privacy service. Readers looking specifically for the China and South Korea argument can skip to 23:39; most of the preceding essay builds the financial machinery behind it.
Published on Trari
Where next?
A professional in this field? Want to write about Money & self-improvement on Trari? Tell us ›
What do you think?
Want to know when somebody answers you?
- Channel
- How Money Works
- Subscribers
- 1,760,000
- Video published
- November 10, 2026
- Video views so far
- 2,155,944 (6 days online)
The numbers
The essay has passed two million views since publication. Here is its audience count alongside the channel's wider totals.
💡 Test your knowledge
Which hardware does the essay describe CoreWeave using as loan collateral?
The essay uses CoreWeave to explain borrowing secured against GPUs, or graphics processing units.

