Who is on each side · Part 08
Who writes the checks for the buildout?
13 voices who disagree about AI Lenders Are Financing the End of Scarcity. Each voice is attributed and paraphrased, dated where the date is known, with a link where one exists.
The DeepStack callSkeptical
Lenders are right to finance capacity and wrong to assume it stays scarce. Scarcity is a season; debt is a contract.
Our Call takes the side “The cash has to triple”.
6 voices
The cash has to triple
The tech silo is betting on a future in which demand for AI and tech services explodes, while the silos covering the companies that would pay for those services see a much more modest outlook, and both cannot be right at the same time.
In S&P 500 Q2 earnings season data, AI adoption keeps broadening: 69% of companies now point to a live deployment, up from 64% last quarter.
Alibaba's shift to financing an AI data center buildout through share issuance is a demerit against the investment thesis, as the buildout may be 'practically interminable' without a clear productive path.
Most analysts covering the former software/social-media/AdTech companies now recast as hyperscalers have never had to model maintenance capital expenditure, and most still are not doing it, or not doing it well.
Using a model assuming equipment costs rise to $100 billion per GW by 2030, five-year equipment cost recovery, 25-year site cost recovery, a 15% annual return on invested capital, and a 65% cloud operating margin, AI labs would need to reach $6.8…
Daniel Romero states they are not arguing the buildout is unsustainable, but that one-year compute paybacks will probably become the exception rather than a baseline assumption, possibly dependent on specific timing and shortages, and that 2026…
5 voices
The spending pays
Based on their capex sustainability model, Market Sentiment estimates that next year's ~$1.2T+ of AI capex should drive ~$2T of incremental revenue, which they say will only be a fraction of the total incremental revenue AI generates as a business…
Aggressive capital spending tends to be suboptimal because after an initial euphoric rise in stock prices tied to rising capex, investors realize they overpaid for the underlying inputs, from land to memory to fiber-optic cables.
The customer base is broadening beyond a concentrated set of hyperscalers to enterprises, AI-native companies and cloud providers, with AI-related infrastructure commitments expected to exceed $3.1 trillion, making the next phase of demand less…
Hyperscaler capex is durable rather than at risk of a sudden cut, because the largest spenders keep reaffirming the necessity of AI investment and their EBITDA and EBIT margins are still rising, giving them capacity to keep funding it.
AWS has had more demand than supply, with customers wanting more AWS capacity for AI than could be served despite large Nvidia orders and Amazon's own Trainium chips, and Kris considers this kind of capacity-constrained spending among the best…
Also in dispute · 6 voices
Cash, or a loop?
2 voices
A closed loop
Over $1T of GPUs bought to win two unprofitable labs, the source of about 70% of hyperscaler AI revenue.
George Noble
Nvidia now gives a year to pay.
4 voices
Real cash, real use
Today's biggest AI infrastructure spenders are large, highly profitable companies like Microsoft, Google, Amazon and Meta, with Microsoft generating more than $55 billion of operating cash flow in its latest quarter and Alphabet roughly $39…
Moving beyond the hyperscalers, financing becomes messier: neoclouds are borrowing enormous amounts of money, AI labs need compute without necessarily having the cash flow to fund it, NVIDIA is investing throughout the ecosystem, infrastructure…
In the August 14, 2026 edition, the author argues: Ren warns the whole theme is AI with more leverage and less liquidity: every fund is under four months old with no track record, the largest is run by a first-time adviser, in a capex pause the…
AGI is not the investable trade; compute is, because Astra adds a second, inference-driven demand chain (more capable models, more agents, more delegated work, longer inference trajectories, more tokens, more compute) alongside the existing…
Also in dispute · 8 voices
What are the lenders holding?
2 voices
Scarcity as collateral
Contracted GPU lease rates and second-hand prices have held up even for chips launched 3 to 6 years ago, and compute is now about 60% of hyperscaler capex, so the scarce, short-lived asset is a growing share of what is being financed.
The entire AI buildout has been constrained by scarcity of power, GPUs, transformers, interconnection, skilled labor and ready-to-use capacity, and this scarcity has propped up the economics of everything financed against it by holding up lease…
6 voices
Guarantees, not faith
Across the five hyperscalers, uncommenced, non-cancelable data-center leases and purchase commitments together total roughly $2.7 trillion in off-balance-sheet contractual commitments as of the disclosed filing dates, with the total potentially…
Meta's data center JV structure shows that the parties closest to the assets do not believe the value will hold: lenders would not bear residual value risk without protection, so Meta issued a residual value guarantee of up to $28 billion on…
George Noble's bear case for AI rests on the question of actual return on investment; he argues that Nvidia announcing massive funding packages just to keep customers like OpenAI able to keep buying chips shows the money underneath the boom is not…
George Noble cites an estimate from Julien Garran at MacroStrategy putting AI malinvestment at roughly twenty four times what was seen during the dot com bubble.
Oracle's free cash flow was positive every year from FY2018 to FY2024, then turned slightly negative in FY2025 (-0.39B), dropped to -23.69B in FY2026 on 55.66B of cash capex, and sits at -28.72B on a trailing twelve months basis through Q1 FY2027…
Oracle's record Q1 operating cash flow of 23.103B (up 184% YoY) is largely explained by 11.363B of customer prepayments with a significant financing component; excluding that prepayment, operating cash flow would be only about 11.740B (up a more…
Notes
Not a disagreement.
Where two voices only seem to differ, or where the thread runs back to an earlier Part.
The price rises with distance from the hyperscaler
The core: investment-grade spread 83bp, bank default swaps flat near 40bp. One step out: guaranteed debt pays 100 to 150bp over the guarantor. The edge: Oracle CDS at a record, CCC spread 1,146bp.
Where they cross
Who answers whom.
The specific points where one voice meets another: the same evidence read two ways, or the same mechanism with a different sign.
Steve Eisman contradicts Gaetano, on the mechanism
Steve Eisman qualifies JunkBondInvestor, on a condition
Torsten Slok refutes Michael Burry, on the evidence
Torsten Slok contradicts Aurelion Research, on premise
Gaetano qualifies Torsten Slok, on the evidence
Daniel Romero qualifies Kris, on a condition
JunkBondInvestor contradicts Market Sentiment, on the evidence
Torsten Slok qualifies Aurelion Research, on the evidence
Kakashii qualifies Aurelion Research, on a condition
Daniel Romero qualifies Michael Burry, on the evidence
Daniel Romero contradicts Aurelion Research, on the evidence
Torsten Slok qualifies Ren, on the mechanism
Gaetano qualifies JunkBondInvestor, on a condition
Gaetano contradicts Steve Eisman, on the evidence
Gaetano qualifies Michael Burry, on the evidence
Kris contradicts Daniel Romero, on the evidence
Torsten Slok qualifies Kakashii, on premise
Daniel Romero qualifies Torsten Slok, on the evidence
Gaetano qualifies Market Sentiment, on the mechanism
Daniel Romero qualifies Torsten Slok, on the mechanism
Gaetano qualifies Daniel Romero, on the evidence
Daniel Romero qualifies JunkBondInvestor, on a condition
Michael Burry contradicts Gaetano, on the evidence
JunkBondInvestor qualifies Torsten Slok, on a condition
What would settle it
The dated tests.
The same tests the story set, on the Docket; results land on the Results page.
September-quarter results test cash flow toward $2 trillion and whether maintenance costs enter accounting.
Part 08: AI Lenders Are Financing the End of Scarcity Pending
New Mexico rules on the fuel-cell permit; approval supports schedule, denial turns Oracle’s notice into a warning.
Part 08: AI Lenders Are Financing the End of Scarcity Pending
Pipeline in-service date tests whether Project Jupiter can secure fuel for its planned campus.
Part 08: AI Lenders Are Financing the End of Scarcity Pending
- First Blackwell renewal
Contract repricing shows whether scarce compute retains value or lenders face faster obsolescence.
Part 08: AI Lenders Are Financing the End of Scarcity Pending
- Fluidstack terms
Government-loan terms reveal the price Washington assigns to financing bottlenecks it wants to remove.
Part 08: AI Lenders Are Financing the End of Scarcity Pending
Analysis and opinion, not investment advice. Voices are paraphrased from public writing and attributed by name; DeepStack shows no score, ranking or accuracy for any person.