AI Credit: Heads I Win, Tails You Lose
A railroad is like a lie – you have to keep building to it to make it stand.
- Mark Twain, 1867
August 2026
In 605 AD, Emperor Yang of Sui drove his empire into the ground building the Grand Canal. The massive waterway became the backbone of the Chinese economy for a thousand years. The Tang inherited the canal and got the golden age.
Between 1866 and 1873, Americans laid 35,000 miles of new railway track. The rails transformed the continent. Banks financing the expansion with debt like Jay Cooke & Co went bankrupt, triggering the Panic of 1873 and a multi-year depression.
Egypt finished building the Suez Canal in 1869. The passage cut travel times between Europe and Asia, revolutionising global shipping. Yet a debt crisis in 1875 forced the Egyptian government to cede its financial control of the canal to the British.
History is not short of examples where worthy projects ended up not being worthy investments. These projects benefited the wider population later on but ended up bringing insufficient financial rewards to their initial capital providers, especially when capital was in the form of credit.
At nearly three percent of GDP, the current AI capex cycle is the second largest in the history of the United States, second only to the annexation of Louisiana in 1803. There are no clear winners and losers yet – and we believe credit investors are exposed to downside from losers, with limited upside from the winners.
We are not sceptical of AI as a technology – we are users ourselves.
However, we are sceptical of the returns to credit providers. Credit investors in AI-linked firms enjoy a capped return on capital, are lending at record low spreads, are exposed to uncertain ROI per dollar and do not share the convex upside of equity investors.
A large part of AI credit comes from private markets – over the past decade private credit as an asset class nearly quadrupled in size, with over 40% of exposure concentrated in tech.
Private credit has been an ingenious way to keep borrowing off balance sheet – benefiting both borrowers as well as investors, who enjoyed returns without mark-to-market. But the tide is turning. A more mature AI cycle will likely differentiate the next Yahoo or Netscape. Credit conditions have never been better: yet, inflation remains persistent, and central banks are less willing to ease. Finally, government spending is finally starting to make bond markets uncomfortable: rising long-end yields might constrain future leverage. Private credit investors have little to gain, and a lot to lose.