Quoth the Raven
· QTR’s Fringe Finance
· August 17, 2026 at 09:46
· ⏱ 27 min read
| Read on Substack ↗
Summary
The article argues that AI is genuinely transformative but that the trillions in AI infrastructure spending have become a circular, reflexive financing bubble that cannot yet be justified by end-user profits. With long-term bond yields surging, household debt at records, and private asset marks untested, the author expects the AI boom to begin crashing in late 2026 or early 2027. For markets, this means a potential violent deleveraging that starts in AI-linked public equities and then spreads to private credit, real estate, and crypto.
•Japan's 10-year JGB hit 2.93% — highest since 1996 — while the U.S. 30-year Treasury hit 5.27%, a level not seen since 2007, in the 24 hours before publication.
•U.S. household debt reached roughly $18.8T at end-June 2026, up ~$400B YoY; credit-card balances were $1.26T and auto loans $1.71T, with 4.7% of total household debt in some delinquency stage.
•Goldman Sachs predicts hyperscaler AI capex could reach $1.4T in 2027; Morgan Stanley estimates Nvidia and the same hyperscalers have ~$1.8T in off-balance-sheet commitments, including $982B in purchase obligations and $822B in leases not yet commenced.
•Nvidia plans to invest up to $100B in OpenAI as OpenAI builds at least 10GW of Nvidia-powered infrastructure; Nvidia also invested $2B in CoreWeave and agreed to buy up to $6.3B of CoreWeave capacity that goes unsold.
•Nvidia is working with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR on platforms targeting more than $500B of AI-infrastructure financing, with an option to backstop up to $125B.
•The author describes AI financing as circular: high valuations fund compute purchases, compute purchases become semiconductor revenue, and that revenue supports further valuations, financing, and orders.
•The internet precedent is explicit: transformative technology and massive overinvestment coexist; the Nasdaq still fell roughly 80% after 1999 even though the internet exceeded predictions.
The author identifies Nvidia as the 'most important cog in AI's circular financing machine' because it finances customers that then buy its GPUs, making its revenue dependent on the same capital it he
The author identifies Nvidia as the 'most important cog in AI's circular financing machine' because it finances customers that then buy its GPUs, making its revenue dependent on the same capital it helps circulate; if AI demand disappoints, earnings and multiple can compress together.
Risk: Nvidia's option to backstop up to $125B of AI-infrastructure financing could create credit losses or force additional capital deployment during an unwind.
CoreWeave is named as a major GPU customer that received $2B from Nvidia and has Nvidia committed to buying up to $6.3B of its unsold capacity, an explicit example of circular AI financing that would
CoreWeave is named as a major GPU customer that received $2B from Nvidia and has Nvidia committed to buying up to $6.3B of its unsold capacity, an explicit example of circular AI financing that would reverse if rental/compute demand slows.
Risk: CoreWeave's high leverage and dependence on Nvidia for both chips and backstopped demand create refinancing and revenue risk.
The article states a 'meaningful decline in the AI complex' would begin as a semiconductor selloff and then spread, making an AI-heavy semiconductor ETF an implied loser from the circular financing un
The article states a 'meaningful decline in the AI complex' would begin as a semiconductor selloff and then spread, making an AI-heavy semiconductor ETF an implied loser from the circular financing unwind even though the technology itself may keep improving.
Risk: The ETF is concentrated in AI winners with high multiples; an AI capex cut or multiple compression would hit the index broadly.
This newsletter, published August 17, 2026,
features Quoth the Raven
discussing NVDA, CRWV, SMH.
3 trade ideas extracted by AI with direction and confidence scoring.