English edition ·
// l0g investigation · chips, credit and power
The Nvidia System
Chips, credit and power at the heart of AI
Twelve instalments published from November 9 to November 15, 2026 form an investigation into industrial power, credit and politics. Their order preserves the development of the reporting and its questions.
- 12
- complete analyses
- 26
- infographics
- EPUB 3
- offline reading
01 / introduction
A chip at the centre of the commitments
Buying a chip is only the beginning. It needs power, a building and customers willing to pay for its work. At the scale of a large data centre, those requirements set off a chain of decisions: secure an electricity connection, finance servers, reserve capacity and sell compute. Nvidia occupies several positions in that chain. It supplies the hardware, invests in selected partners and guarantees obligations that help some projects raise money. Its influence reaches far beyond processor performance. [B06] [B16]
The crisis surrounding Jensen Huang brings those connections into sharper focus. A boardroom decision can alter a lender’s confidence. An investigation into the destination of machines can raise the stakes of export licensing. A delayed power connection can push back a campus’s expected revenue and debt repayments. These links give the affair its industrial, financial and political reach. Understanding them requires attention to each party’s obligations, dates and conditions.
An original introduction opens the book and connects the twelve analyses.
Sources for the synthesis: SEC · Nvidia, 8-K · Federal Reserve · Financial Stability Report.
02 / reading journey
A crisis traced through twelve analyses
The twelve complete analyses, in publication order. Each chapter remains available online with its references and limitations.
01Nvidia, Trump: a warning at the top
Nvidia, Trump and Megaspeed: espionage warnings, conflicts of interest, AI financing and the risks to US national security.
02Nvidia, Trump and Lutnick: national security under pressure
Warnings, the Megaspeed inquiry, H200s in Xi’an and ballroom funding: Washington’s decisions and Nvidia’s interests put independent enforcement under scrutiny.
03Nvidia’s board faces the Huang problem
An alleged NDA, a shareholder lawsuit and AI financing: Nvidia’s board faces a test of control as a federal equity stake remains a political scenario.
04Nvidia after Huang: Washington faces questions
Huang’s exit, Singapore’s requests for evidence and Nvidia’s succession reshape the debate over US influence and AI financing.
05Nvidia: Trump backs Tim Cook as Lutnick faces pressure
Trump proposes Tim Cook for Nvidia as Lutnick faces scrutiny. The legal, financial and industrial routes to US influence deserve closer examination.
06Nvidia: Congress follows the export-control decision trail
Congress seeks answers from Nvidia’s lawyer and a top export official as Lutnick agrees to testify and Trump’s support for Huang emerges.
07Nvidia: the guarantees behind AI’s billions
Nvidia faces demands for stronger guarantees as Singapore and Malaysia sharpen their scrutiny of server deals.
08Nvidia: Shanghai’s pledged servers and Washington’s blame game
Servers pledged in Shanghai expose questions for AI lenders, while Bessent points to Commerce and Republicans demand answers from Howard Lutnick.
09Nvidia: Chinese AI credit meets US export controls
Chinese AI projects seek bank funding as US export rules shape delivery and the Haas report brings SoftBank ties into focus.
10Nvidia: AI guarantees put the whole credit chain at stake
The Fed maps AI financing risks as Nscale commits to equipment orders while still seeking funding for its compute contracts.
11Nvidia: Huang’s assurances face an independent inquiry
Reuters reports Huang misled Nvidia leaders as the board investigates, Congress seeks testimony and a Speedmatrix land offer expires.
12Nvidia after Huang: contracts, credit and the cost of the crisis
Nvidia has funded Naver. Nscale’s next billion is pending. Contracts, guarantees, lawsuits and public demand reveal the stakes for Nvidia.
03 / reference
The language of credit and risk
A glossary accompanies the book. These reference pages also support further reading online.
- GPU
- A processor designed for parallel computation. A model name, a chip count and a server count describe different things.
- Special-purpose vehicle
- An entity set up to hold a project, its assets and contracts. Agreements determine which assets and parties creditors can claim against.
- Credit enhancement
- Protection intended to improve a lender’s position, such as a guarantee, reserve or subordinated contribution. Its cost and beneficiary matter alongside its amount.
- Residual value guarantee
- Contractual protection based on an agreed minimum value for an asset or lease. The guarantor covers a defined shortfall, subject to the agreed conditions and recoveries. The contract specifies the beneficiary, cap and trigger.
- Collateral
- An asset pledged to secure an obligation. Its stated value, the loan principal and the amount recoverable after default require separate examination.
- Debt-service coverage ratio
- Cash available for debt service divided by interest and principal payments due over the same period. Agreements determine the precise calculation. Lower receipts or delayed commissioning can weaken coverage.
04 / conclusion
AI financing enters the political arena
The affair is far from over. Huang’s departure changed Nvidia’s leadership while the board inquiry, legal proceedings and requests for explanations continued on separate tracks. The next stage also lies in the execution of commitments: funding made available, servers accepted, revenue collected and guarantees that may be called. Those transactions will help establish the crisis’s economic significance. [B01] [B03]
Nvidia sits at the centre of these relationships. The AI financing atlas shows the group acting as supplier, investor and guarantor of selected obligations. Capital helps customers order equipment. That equipment must then generate enough revenue to cover power, rent and debt service. One company therefore occupies several positions in a network whose resilience depends on the future business of data centres. [B06] [B16]
The full conclusion turns to political accountability, AI financing and the decisions still ahead.
05 / sources and edition
Sources, method and limits
Edition dated 15 November 2026. Documentary cut-off: 15 November 2026.
This edition was assembled on November 15, 2026 from the twelve instalments published between November 9 and November 15. The primary-source status check cited in the conclusion took place on November 15 at 12:31 UTC, or 13:31 in Paris. The colophon records this edition’s editorial freeze timestamp. Events, source publication and the book’s public release have separate timestamps.
Text, contents, glossary and infographics are available offline. External sources require a connection. Image enlargement and displayed colours depend on your reading app.
The covers and panorama are conceptual illustrations created with AI assistance. The infographics retain the articles’ dates, units, sources and limitations.
The book is free and available as a direct download. Reuse of original text and infographics: Creative Commons Attribution 4.0 International.
To help keep these investigations freely available, the support page accepts one-time or monthly contributions, with Bitcoin support also available. Support l0g
Check file integrity
EPUB 3 · 13.1 MB · SHA-256
25cbf85336afd0a096389fccafd882dad3cdae1d59ef46be09990defdb3d5ae7