Alphabet’s filings show that its backstops for other companies’ data-center payment obligations had a notional value of $43.8 billion on 30 June 2026, up from $0 on 31 December 2024. Alphabet says the notional amount represents “the maximum potential exposure regarding future payments in the event of specified default scenarios by underlying parties”.
By our arithmetic, that is 6.7 times the $6.5 billion Alphabet reported nine months earlier, on 30 September 2025. Its third-quarter 2026 results release of 28 October 2026 does not update the figure, and its third-quarter 10-Q had not been filed when we checked SEC EDGAR.
A vertical news Short narrated from the published headlines and opening lines of this piece and four related FourWeekMBA pieces.
Business Pill · Free Cash Flow
An evergreen Business Engineer explainer on free cash flow, the cash a company has left after capital expenditures. It explains the general idea only and does not discuss Alphabet or its backstops.
The key insight: By our arithmetic from Alphabet’s filings, its data-center backstops grew 6.7 times in nine months, and with its energy guarantees came to $51.4 billion, equal to 52% of its long-term debt on 30 June 2026.
Alphabet’s first-quarter 2026 10-Q says: “We have entered into agreements with certain third parties to backstop certain payment obligations relating to data centers, which we account for as credit derivatives”.
The same filing says the agreements “carry remaining terms of up to 15 years and the total potential exposure reduces over time as the underlying parties fulfill their payment obligations”. Its 10-Q for the third quarter of 2025 had described remaining agreement periods of up to 10 years.
On a default, Alphabet says it keeps “the right to assume the underlying leases for internal use or to sublease to third parties”. It adds that it may end a backstop by making a termination payment, and that such payments “may be partially offset by equity or cash receipts from counterparties”, which are not reflected in the notional amounts.
Notional amount of Alphabet’s credit derivatives backstopping third-party data-center payment obligations, which Alphabet says represents the maximum potential exposure in specified default scenarios, $ billions, 31 December 2024 to 30 June 2026. Excludes $7.6 billion of energy-equipment financial guarantees at 30 June 2026. The 6.7x multiple is our arithmetic. Sources: Alphabet Forms 10-Q and 10-K.
The Link to TPU Sales
In its first-quarter 2026 10-Q, Alphabet says Google Cloud “has entered into a limited number of agreements to supply multiple gigawatts of TPU hardware to customers who require or provide on-premises infrastructure for specialized, high-scale workloads”. The same passage adds: “In connection with certain of these agreements, we have agreed to provide credit backstops to support third-party data centers and power infrastructure”.
The filings do not say how much of the $43.8 billion relates to those TPU agreements.
Quarter by Quarter
The notional amount was $0 on 31 December 2024, $6.5 billion on 30 September 2025, $16.9 billion on 31 December 2025, $28.4 billion on 31 March 2026 and $43.8 billion on 30 June 2026, according to Alphabet’s 10-Qs and its 2025 10-K.
The first-quarter 10-Q also says that in April 2026 Alphabet entered into additional backstop agreements with notional amounts of approximately $15.3 billion. By our arithmetic, the second-quarter increase was $15.3 billion.
The fair value of the credit-derivative liabilities was $69 million on 31 December 2025, $339 million on 31 March 2026 and $815 million on 30 June 2026, according to the same filings.
The Energy Guarantees Alongside
Separately, Alphabet reports financial guarantees that “support counterparty procurement of long-lead time equipment for our future power purchase and energy agreements”. Their maximum potential future payments were $5.7 billion on 31 December 2025, $9.0 billion on 31 March 2026 and $7.6 billion on 30 June 2026.
By our arithmetic, the data-center backstops and the energy guarantees together came to $51.4 billion on 30 June 2026, equal to 52% of Alphabet’s long-term debt of $98.2 billion on the same date.
As we read the filings, the data centers themselves stay off Alphabet’s balance sheet: Alphabet says it does not consolidate these entities because it does not direct the activities that most significantly impact their economic performance.
The filings tie part of this to TPU sales: Alphabet says that in connection with certain of its on-premises TPU supply agreements it agreed to provide credit backstops to support third-party data centers and power infrastructure.
The fair value Alphabet records for the credit-derivative liabilities rose from $69 million to $815 million in the first half of 2026, against a notional amount of $43.8 billion.
Alphabet, Form 10-K for 2025, risk factors
“leases, backstops, guarantees, and potential liabilities from large commercial agreements, could harm our financial condition and reduce our financial and business flexibility”
Three Implications
FOR GOOGLE CLOUD Alphabet’s 10-Q links credit backstops to certain of its agreements to supply multiple gigawatts of TPU hardware for on-premises use.
FOR THE BALANCE SHEET Alphabet says the notional amount is its maximum potential exposure in specified default scenarios, and that it may assume or sublease the underlying leases after a default.
FOR TIMING Alphabet says the agreements carry remaining terms of up to 15 years and that the exposure reduces as the underlying parties pay.
The framework argues that the AI build-out “is a sovereign-scale financing problem dressed in tech-company clothing — and the financial architecture is starting to invert.”
Alphabet’s filings show one form that architecture takes: $43.8 billion of promises to cover other companies’ data-center payments, recorded as credit derivatives rather than debt. The framework does not discuss these figures.
What Is Not Established
The filings do not name the counterparties, the data centers or the lenders whose payments are backstopped, and do not split the $43.8 billion between the TPU agreements and other arrangements.
A notional amount is a maximum under specified default scenarios, not an expected loss. Our 52% compares a contingent maximum with debt on Alphabet’s balance sheet; it is a scale comparison, not a measure of risk.
This Week In The AI Supercycle
Execution Fault Lines, Concentrated Demand, and Geopolitical Firewalls
This week’s issue: the AI build is not short of demand for its parts. It is short of cheap money, and its end revenue rests on a narrow base of large, often venture-funded buyers.
Alphabet’s filings record a new kind of exposure that grew from nothing to $43.8 billion in 18 months: promises to cover other companies’ data-center payments if they default, with terms of up to 15 years and, in some cases, tied to its TPU supply agreements.
A note on sourcing. Figures come from Alphabet’s Forms 10-Q for the quarters ended 30 September 2025, 31 March 2026 and 30 June 2026 and its 2025 Form 10-K, all filed with the SEC. The multiple, the sum and the percentage are our arithmetic; linked earlier pieces carry their own sources.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
