Observed Signal · Sep 28, 2026 · Financial Analysis · Source: CNBC Investing · Impact: 3/5 · Sentiment: Negative
AI companies adopt risky 'take-or-pay' contracts for compute
The AI industry is increasingly relying on take-or-pay contracts to secure computing power, a model borrowed from the energy sector. These agreements require AI companies to pay for GPUs, server time, and other inputs regardless of usage, creating significant financial risks. The payments are backed by expected future revenue, but there are concerns about whether AI revenues will materialize in time to meet payment schedules. Off-balance-sheet commitments are growing, with estimates ranging from $1.65 trillion to $3 trillion. Major players like Anthropic and OpenAI have committed billions, while companies like CoreWeave and Alphabet rely heavily on such contracts. Financial analysts and regulators warn that the entire system could collapse if companies fail to secure enough capital. The Bank of International Settlements has highlighted potential systemic risks, particularly through special purpose vehicles. The article notes that major banks and investors are closely monitoring this trend, with experts predicting a 'wave of disputes' as the sector matures.
The article highlights significant financial risks in the AI infrastructure buildout due to take-or-pay contracts, which could impact the broader tech and cloud industry, including advertising tech that relies on AI capabilities.
Track Anthropic Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your free Explorer workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Take-or-pay contracts obligate AI companies to pay for compute regardless of usage, similar to energy sector agreements.
- Anthropic committed $750 billion for compute through 2030; OpenAI could spend over $500 billion in the next 10 years.
- Off-balance-sheet commitments in tech sector estimated at $1.65 trillion to $3 trillion.
- Akamai and Anthropic agreed to a take-or-pay contract on Friday (Sept 28, 2026).
- CoreWeave gets majority of revenue from take-or-pay contracts, and Alphabet lists them in $811 billion purchase obligations.
Connected Companies & Entities
9 Entities mapped“Anthropic has committed $750 billion for compute through 2030, according to the Journal....”
“Take-or-pay contracts, such as the one agreed to Friday between Akamai and Anthropic, are agreements that frontier labs sign with cloud plat...”
“OpenAI could spend more than $500 billion for various inputs in the next 10 years, The Information reported....”
“Neocloud CoreWeave gets the majority of its revenue from take-or-pays, according to an April 15 note from Morgan Stanley....”
“Alphabet lists take-or-pays as a piece of its $811 billion in purchase obligations....”
“Citizens Securities said in a Sept. 17 note that it's unsure where Oracle will get the money for promised building projects aimed at reachin...”
“Nvidia, which reported a 63% surge in accounts receivable during the first half of this year....”
“Meta Platforms, for example, will lease the Hyperion data center in Louisiana from such a special purpose vehicle....”
“CoreWeave gets the majority of its revenue from take-or-pays, according to an April 15 note from Morgan Stanley....”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
AI Data Center Boom 'Stress‑Tests' Insurers and Lenders
Rapid private-capital financing and large-scale deals for AI data centers are creating capacity and risk strains for insurers and lenders. McKinsey estimates global data-center spending could reach $7 trillion by 2030, and Preqin data show private infrastructure data-center deals were consistently above $10 billion last year, including a $40 billion consortium purchase of Aligned Data Centers involving Nvidia, Microsoft, BlackRock and xAI. Insurers say concentrating tens of billions in single campuses has created capacity and underwriting challenges, while new financing structures—such as GPU-backed loans—raise lifecycle and opacity risks. CoreWeave announced an $8.5 billion investment-grade GPU-backed deal. Industry participants (insurers, brokers, law firms, and senators) have raised concerns about off‑balance-sheet financing, concentration risk, and potential second‑order litigation for downstream investors.
AI Industry Fuels Itself with Circular Cloud Deals
This analysis argues the modern AI industry is sustained by large, circular financing and cloud-spend commitments between AI labs (notably Anthropic and OpenAI) and hyperscale cloud providers (Google, Amazon, Microsoft, Oracle, Nvidia). Anthropic and OpenAI have each committed hundreds of billions in future cloud spending, which hyperscalers book as revenue backlog while also investing equity in those labs. The author warns this circularity depends on theoretical capability translating into reliable, widely adopted products; but recent research and usage metrics show capability is improving faster than real-world reliability and adoption. The piece concludes the industry's stability hinges on run-rate revenue converting into durable, scaled customer value rather than contingent, circular commitments.
AI Must Generate $3 Trillion to Justify Infrastructure
An analysis traces escalating AI infrastructure costs and the revenue required to justify them. Sequoia partner David Cahn updated his 2023 model and estimates $1.5 trillion in AI infrastructure spending for 2026, concluding the AI industry must earn roughly $3 trillion to pay back chips and data-center expenditures. Major model makers show large revenues (Anthropic ~ $60B ARR; OpenAI reported $13B in 2025 and previously claimed $20B ARR in Nov 2025), but a substantial gap remains. Apollo economist Torsten Slok warns hyperscalers (Google, Meta, Microsoft, Amazon) expect big free-cash-flow improvements by 2028 and that failure to meet those targets could trigger severe market reactions. Downward pressures include the rise of cheaper open-weight models and falling token prices; OpenAI’s latest model is cited as 54% more token-efficient on coding tasks.
Track Real-Time Market Signals & Shifts
Set up custom watchlists to receive automated, evidence-grounded executive digests whenever material signals or shifts occur across your tracked landscape.