Risky Financing of AI Infrastructure and the Role of Google/Nvidia

Carte mentaleMind map

The article compares the current frenzy of AI investment to 19th-century railway speculative bubbles, notably the Panic of 1873 triggered by Jay Cooke's risky financing. It highlights that major tech companies, with the notable exception of Microsoft, are taking on massive debt to fund their AI infrastructure, with Google even issuing equity.

It sheds light on Google's strategy, which, despite internal challenges at DeepMind, positions itself to dominate AI infrastructure by selling its computing capacity (TPUs) to major players like Anthropic. This approach, compared to Berkshire Hathaway's in railways, aims to generate massive profits as a platform, even if it means funding competitors.

Finally, the article analyzes Nvidia's attempt to transform its AI factories into a new class of investable assets, by partnering with large asset managers to raise capital. This initiative, while innovative, is perceived as risky, as Nvidia must guarantee a portion of the residual value, which could be a sign of pressure on its margins in the face of competition from alternative chips and the potential commoditization of intelligence.