We Asked Andy Constan What Happens If AI Funding Breaks Before the Thesis — And if Warsh Blinks

Episode Summary

Executive Summary: Andy Conston argues that today’s market regime is internally consistent: strong productivity-led growth, heavy Treasury/corporate issuance, and persistent policy support explain high stocks and high rates. He says inflation won’t return to target unless policymakers let asset prices fall, and warns that AI capex will increasingly depend on capital markets absorbing massive funding needs, which could become the next market stress point.

Main Topics: Why stocks and yields can both be high (Priority: 5/5): Conston explains that stronger real growth, especially productivity growth tied to AI/data centers, naturally pushes real yields higher while supporting equities because investors prefer growth assets over bonds. Inflation and the limits of short-rate policy (Priority: 5/5): He argues the current economy is not a classic debt cycle, so Fed short-term rates are less effective at cooling demand. In his view, reducing inflation requires a fall in asset prices and the resulting loss of wealth effect. Yield suppression and long-end support (Priority: 5/5): The discussion details how policymakers have supported long-term bonds through balance-sheet policy, bill-heavy Treasury financing, regulatory changes, and other interventions to avoid market stress and financial instability. Warsh, the Fed, and a potential policy shift (Priority: 4/5): Conston sees early signs that Warsh may allow more long-end rate pressure and reduce bond-market suppression through balance sheet changes, but says this is only a 'drumbeat' and not yet confirmed. Treasury issuance and quarterly refunding (Priority: 4/5): He describes how auction composition matters: more bills can support risk assets, while more coupon issuance is bearish for bonds and equities. Future deficit growth may force larger issuance and tighter financial conditions. AI capex, circular financing, and the 'hamburger thesis' (Priority: 5/5): Conston argues that near-term AI valuation depends less on ROI than on whether capital markets can finance enormous capex and revenue promises. He worries the market may eventually 'close' before all the needed funding is absorbed. NVIDIA, private credit, and new financing structures (Priority: 4/5): He interprets NVIDIA’s large financing/backstop arrangements as evidence that traditional direct funding channels are getting strained, pushing issuers toward more complex structured-credit-like solutions.

Key Arguments: Current market pricing is internally consistent: higher growth expectations and intense issuance explain both strong equities and elevated long-term yields. Real productivity growth increases demand for capital, which pushes real yields higher because investors prefer owning growth assets rather than bonds. The U.S. government debt burden is not, by itself, an imminent crisis because the government can print money and the relevant issue is whether private investments earn above their cost of capital. Inflation is unlikely to return to target through short-term rate policy alone; to reduce demand, policymakers must allow asset prices to fall and wealth effects to reverse. Long-end bond suppression has been achieved through bill-heavy funding, balance-sheet policy, emergency facilities, regulatory changes, and mortgage support; removing these props would be materially tightening. Warsh may be signaling a willingness to let the bond market do some of the Fed’s work, but this is still tentative and could be offset by Treasury choices. AI capex funding is more important in the near term than ROI because the market must first absorb trillions in issuance before eventual returns matter. Capital markets can function smoothly for a while, then suddenly 'close' when investors balk at concessions; that inflection could create a sharp repricing in AI-linked assets. NVIDIA-style structures show that issuers are using increasingly complex financing to keep the AI buildout going, which is supportive of growth but also a sign of strain. The key practical risk is not tomorrow’s news but the accumulation of supply and financing stress that may appear after the market has already priced in optimistic future earnings.

Data Points: U.S. marketable debt: roughly $32 trillion - Conston cites this as the main stock of marketable federal debt Social Security / intra-government debt: roughly $8 trillion - Added to marketable debt in his discussion of total federal obligations Average interest rate on most government debt: about 2.68% to near 3% - He says the weighted average cost of debt remains manageable Bills as share of outstanding debt: about 20% to 21.5% - He uses this to show how Treasury financing has been skewed toward short duration Potential higher interest cost example: $250 billion - Illustrative additional annual interest if rates rose on a large debt base AI capex this year: $600 billion to $700 billion - Estimate of current annual spending on AI infrastructure AI capex next year: $1 trillion - Projected spending if buildout continues Net corporate share reduction historically: about $1 trillion per year - He says U.S. corporations have been net retiring shares through buybacks SPX earnings growth assumption: mid-20s % - He says broad earnings forecasts assume strong AI-driven growth Google stock sale price: 355 - He notes Google issued stock at 355 and it later traded lower Current Google trading level mentioned: 342 - Used to show issuance being underwater SpaceX deal price: 135 - Referenced as a financing comp in the private market 2021-2022 stimulus impact: large income replacement effect - He describes government/fed stimulus as replacing household income rather than just lowering rates Demographic ratio in 1952: 6 workers per senior - Used to illustrate long-run support for economic growth Demographic ratio in 2011: 2.7 workers per senior - Shows a major decline in support ratio Projected demographic ratio by 2030: 2.5 workers per senior - Cited as evidence of slower potential growth

Pivotal Quotes: "the only way you do it is by hitting asset prices" — Andy Conston: Explaining how inflation can be reduced when short-term rates are ineffective "the script is you have to do that" — Andy Conston: Referring to the need for weaker asset prices and loss of wealth effect to tame inflation "The issue is if all this gets through the system, all the issuance, the trillions of dollars... then we can start focusing on the ROI" — Andy Conston: Summarizing why AI funding and capital absorption matter more than AI returns in the near term

Implications: Investors should watch long-end rate policy, Treasury issuance, and AI financing capacity more than headline Fed cuts. The biggest risk is a sudden capital-markets repricing that hits bonds, equities, and AI-linked names at once.