Thesis · AI Financing, Credit & Capital Structure

Will Liberty Energy (LBRT) / Liberty Power Innovations disclose at least $1.5bn of closed non-recourse SPV project debt backed by fully executed energy services agreements by year-end 2027?

Will Liberty disclose at least $1.5bn of closed non-recourse SPV project debt backed by executed hyperscaler ESAs by year-end 2027? Yes side argues the structure is bankable and will close; no side argues it is a vehicle for hiding parent leverage and risk.

6 dated tests · 0 resolved · next on

The tests More theses in AI Financing, Credit & Capital Structure

Who is on each side

The two sides.

Each side in DeepStack’s words, then the voices who argue it, attributed and paraphrased, dated where the date is known. A voice links to its page on DeepStack where one exists.

The case for yes

The SPV structure is bankable and closes against executed ESAs

Investment-grade counterparty ESAs make project-level cash flows bankable, letting lenders underwrite the SPV without recourse to the parent. Management intends to establish special purpose vehicles backed by fully executed ESAs to raise non-recourse debt for project-level capex, keeping roughly $3.2bn of the $4.5bn buildout capex off the parent balance sheet. That ring-fenced, disclosed structure with identified lenders is what closes the debt rather than merely disguising parent leverage.

The case for no

Off-balance-sheet vehicles hide leverage and will not close as claimed

Hyperscalers leasing rather than owning data centers, shifting financing to private-credit-funded shell operators, so AI bubble losses landing on pension and insurance beneficiaries instead of Big Tech balance sheets. Post-Enron rules raise minimum independent equity to 10 percent and use risks-and-rewards test, but structures are designed to comply while retaining economic benefit to parent, so neoclouds and data center SPVs remain off-balance-sheet while Nvidia retains the substantive risk and benefit. It would show first if no SPV debt closes and the buildout is funded with parent debt or equity.

1 voice

The SPV structure is bankable and closes against executed ESAs

  1. JunkBondInvestor

    Management intends to establish special purpose vehicles backed by fully executed ESAs to raise non-recourse debt for project-level capex, keeping approximately $3.2bn of the $4.5bn buildout capex off the parent balance sheet. This structure is…

3 voices

Off-balance-sheet vehicles hide leverage and will not close as claimed

  1. Prof G Markets

    Off-balance-sheet AI leverage routed to household savings

  2. Kakashii

    FASB's 2003 post-Enron reforms increased the independent equity threshold from 3 percent to 10 percent and shifted from voting-control to risks-and-rewards tests for consolidation. Nvidia's structures meet the technical criteria: neoclouds are…

  3. Michael Burry

    A life insurer sells long-duration retirement annuities, which require long-duration assets to offset them, and creates data-center-backed ABS against fifteen-to-nineteen-year lease income. It then sets up a captive reinsurer in Bermuda that it…

What would settle it

The dated tests.

Each test names the public source that settles it, the rule and the day. The Call’s number, the readers’ median and the machines’ numbers stand beside it, as on the Docket.

  1. Fluence Energy, operating lease liabilities growth from a year earlier, quarter ended September 30, 2026: whether it comes in at or above 70%.

    On the Docket Pendingth055-t01Open question

    Specification pending

    The machinesClaude 25% · DeepSeek 78%

    Claude: 25%DeepSeek: 78%

    Your number

  2. Fluence Energy, operating lease liabilities growth from a year earlier, quarter ended September 30, 2026: whether it comes in at or above 85%.

    On the Docket Pendingth055-t02Open question

    Specification pending

    The machinesClaude 15% · DeepSeek 9%

    Claude: 15%DeepSeek: 9%

    Your number

  3. Fluence Energy, operating lease liabilities growth from a year earlier, quarter ended September 30, 2026: whether it comes in at or above 95%.

    On the Docket Pendingth055-t03Open question

    Specification pending

    The machinesClaude 13% · DeepSeek 10%

    Claude: 13%DeepSeek: 10%

    Your number

  4. Oracle, interest expense as a share of revenue, quarter ended November 30, 2026: whether it comes in at or above 7.5%.

    On the Docket Pendingth055-t04Open question

    Specification pending

    The machinesClaude 72% · DeepSeek 72%

    Claude: 72%DeepSeek: 72%

    Your number

  5. Oracle, interest expense as a share of revenue, quarter ended November 30, 2026: whether it comes in at or above 8.2%.

    On the Docket Pendingth055-t05Open question

    Specification pending

    The machinesClaude 61% · DeepSeek 72%

    Claude: 61%DeepSeek: 72%

    Your number

  6. Oracle, interest expense as a share of revenue, quarter ended November 30, 2026: whether it comes in at or above 9%.

    On the Docket Pendingth055-t06Open question

    Specification pending

    The machinesClaude 33% · DeepSeek 12%

    Claude: 33%DeepSeek: 12%

    Your number

In The Buildout

The stories that take up this thesis, with their own Calls and tests.

A thesis page states one question and both sides of it; DeepStack takes no side in words, and its view is the number it gives each test before the date. 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. Pre-specified means the words were approved on or before the test’s date. Specified after the date means they were written afterwards: such a test is counted and shown, and never rated. An open question has no words yet. The Call’s number is the chance, from 0 to 100, that the Call gives the yes answer to a test’s question, recorded before the test’s date; it is DeepStack’s own reading, not a market price and not advice. The readers’ mark on the same line is the readers’ median, the middle of the numbers signed-in readers gave the test, shown from 25 readers. The machines: Each model read this page and nothing else. The number that counts is the latest one each model gave before the test closed.