The US economy is undergoing a profound structural shift, rapidly transitioning from the post-financial crisis era of deleveraging to a debt-fueled growth model, with artificial intelligence at its core.
Strategist Albert Edwards from Societe Generale cautions in his latest weekly report that the simultaneous surge in corporate and household borrowing is pushing the American economy toward a perilous tipping point. According to UBS calculations, the US has added approximately $800 billion in new credit over the past year, accounting for 2.6% of GDP. The US now drives more than half of the global credit impulse, with a significant portion stemming from financing needs for AI-related projects.
Concurrently, the US personal savings rate has plummeted to an unusually low 2.6%, making consumption support increasingly fragile. Edwards states plainly that both consumption and investment are currently highly dependent on the premise that the AI "bubble" does not burst.
The implication for markets is clear: if the AI-driven credit expansion reverses, the world's largest engine for the global credit impulse would suddenly stall, sharply elevating the risk of an economic downturn.
The US Takes the Lead in Global Credit Expansion, AI is the Core Driver
The global credit landscape is undergoing a historic restructuring. According to UBS data, the global credit impulse, measured by the 12-month net new credit as a percentage of GDP, has shifted from being led by other economies to an overwhelmingly "American story." The US contributed roughly $800 billion in incremental credit over the past year, representing more than half of the total global credit impulse.
The primary driver of this credit expansion is AI capital expenditure. Reports indicate the US economy is pivoting away from the long deleveraging cycle following the 2008 financial crisis toward a "more debt-driven" growth model. Morgan Stanley estimates show the total leverage ratio for hyperscale cloud computing companies doubled from 0.9x to 1.8x in just two quarters. Hyperscale cloud companies alone are projected to drive $600 billion in debt issuance by 2026.
Edwards points out that the rapid expansion of AI-related debt is developing characteristics of a standalone debt bubble. He cites detailed data from the latest Federal Reserve Z.1 Financial Accounts report, emphasizing that the trend of surging corporate borrowing is now corroborated by official statistics.
Savings Rate Falls to Dangerous Low, Consumption Support is Precarious
Edwards highlights a risk that has received relatively less market attention: the concurrent surge in US household borrowing. The current US personal savings rate has fallen to 2.6%, an historically anomalous low level that Edwards believes "may be unsustainable."
He likens the current situation of the American consumer to the cartoon image of Wile E. Coyote running off a cliff edge and briefly suspended in mid-air—apparently still functioning, but with no real support. More alarmingly, key indicators measuring real household income have already shown year-on-year contraction.
Edwards's logic chain is clear: the reason consumers can maintain roughly 2% consumption growth despite declining incomes is fundamentally due to the "feeling of being wealthier" from stock market wealth effects, which suppresses the willingness to save. However, should asset prices decline, the savings rate would be forced to rebound—the precedent being the post-2020 COVID shock when the savings rate briefly soared above 30%. At that point, consumption would contract in line with income, and the economic impact would far exceed current market expectations.
Liquidity Shifts from Financial Markets to Real Economy, Pressuring Asset Valuations
Edwards issues a more direct warning for investors: the simultaneous leveraging by both corporations and households means a significant amount of liquidity is being "siphoned" from financial markets into the real economy, posing a threat to currently elevated asset valuations.
He views this process as essentially a reversal of the "secular stagnation" theme, where savings chronically exceed investment. Edwards speculates that the recent sharp declines in precious metals and cryptocurrencies might be early signals of tightening financial market liquidity.
From a macro-monetary perspective, the rapid growth in broad money supply also harbors inflationary pressures, adding further uncertainty to the Federal Reserve's policy path.
Economic Fate Hinges on AI, Debt Efficiency Shows Underlying Concerns
Edwards's conclusion aligns closely with recent analyses from various quarters: both US consumption and investment are currently predicated on the assumption that the AI bubble does not burst. AI has not only captured the stock market—the entire year-to-date gains of the S&P 500 are attributable to its ten largest companies—but has also deeply embedded itself in the real economy's credit cycle.
Data also reveals another layer of concern: although the US has become the dominant force in global credit expansion, it now requires substantial debt to generate a unit of GDP growth, indicating declining marginal efficiency of credit expansion. This implies that if investors begin to question the commercial return prospects of AI, the entire debt-driven growth logic will face systemic pressure.
Edwards concludes with a succinct warning: "Keep a close eye on this heavily indebted area."
Comments