Business Pill · Paying for AI chips whether you use them or not
A take-or-pay contract commits the buyer to pay for a fixed amount whether it uses it or not. The owner gets certainty, and the buyer gets guaranteed capacity and a fixed bill.
A short explainer video, under a minute. The gym, the AI lab and the data centre owner are an illustration.
At a glance
What a Take-or-Pay Contract Is
That is a take-or-pay contract. The buyer commits to pay for a fixed amount, whether it uses it or not.
What Each Side Gets
The owner gets certainty. It can show the contract to a bank, and borrow against it.
Why It Matters
The video turns this into one question: what have we promised to pay for, used or not?
The Short Answer
Think of a gym membership. You pay every month, whether you go or not.
Now make it bigger. In the video, an AI lab signs for a thousand chips, for five years. Used or not, it pays.
What a Take-or-Pay Contract Is
That is a take-or-pay contract. The buyer commits to pay for a fixed amount, whether it uses it or not.
What Each Side Gets
The owner gets certainty. It can show the contract to a bank, and borrow against it.
The lab gets guaranteed capacity, and a fixed bill, against revenue that is not fixed at all.
The Idea in Parts
The lesson in its own parts
What a Take-or-Pay Contract Is
That is a take-or-pay contract. The buyer commits to pay for a fixed amount, whether it uses it or not.
What Each Side Gets
The owner gets certainty. It can show the contract to a bank, and borrow against it.
When Demand Slows
If demand slows, the chips sit idle, and the invoice still arrives.
When Demand Slows
If demand slows, the chips sit idle, and the invoice still arrives.
Why It Matters
The video turns this into one question: what have we promised to pay for, used or not?
A commitment buys certainty for the seller, and a fixed cost for the buyer.
The key insight: A commitment buys certainty for the seller, and a fixed cost for the buyer.
The honest limit
If demand slows, the chips sit idle, and the invoice still arrives.
Where This Shows Up in AI Business Today
From our news coverage
Anthropicโs $518B Leak: Google $111.1B, Amazon $110B, 80% Locked
Commitments that bind whether or not they are used show up in our recent reporting. Our report says that, as Reuters’ account of a confidential prospectus is relayed by ANI, Anthropic expects to spend at least $518 billion over a decade on AI computing from six partners, with about 80% non-cancellable or payable regardless of usage.
From our news coverage
Reflectionโs SpaceX Compute Deal: $150M a Month, 90-Day Exit
In our other report, CNBC reported that Reflection agreed to pay SpaceX $150 million a month from 1 July 2026 through 2029, with either side able to exit on 90 days’ notice after three months. Both reports verified none of the figures.
The Business Engineer Lens
Where this idea sits in the Business Engineer library:
The One Question to Ask
- What have we promised to pay for, used or not?
How This Concept Connects
Layer 1 of 8 · Compute & Energy, on the Business Engineer Concept Map
- Take-or-Pay is an example of Binding Order. Take-or-pay is a binding commitment: the lab pays for the chips whether used or not, so walking away costs something.
- Take-or-Pay contrasts with Usage-Based Pricing. Take-or-pay fixes the buyer’s bill regardless of use; usage-based pricing makes the bill follow the work done.
- Take-or-Pay drives the cost of Burn Rate. A fixed chip bill keeps arriving when demand slows, and the burn lesson says AI burn is often the computing bill.
- Backlog builds on Take-or-Pay. AI backlogs are large because customers sign long contracts to secure scarce capacity, the commitment take-or-pay describes.
- Power Purchase Agreement contrasts with Take-or-Pay. Both are long contracts: a PPA buys a steady power price, while take-or-pay commits payment for capacity used or not.
- Neoclouds (lesson unlocks soon) builds on Take-or-Pay. A neocloud borrows heavily to buy chips; take-or-pay contracts are what let a chip owner show a bank committed payments.
- An Option (lesson unlocks soon) contrasts with Take-or-Pay. Take-or-pay pays whether used or not; an option pays a small fee now and leaves the choice for later.
- Prepayment (lesson unlocks soon) contrasts with Take-or-Pay. Both commit the buyer early: prepayment pays before delivery, take-or-pay pays for capacity whether used or not.
The AI Capex Map
Long compute commitments underpin the capex build-out.Business Engineer framework
The Subsidized AGI Economy
Fixed chip bills against uncertain revenue relate to subsidised AI economics.
Get the full Concept Map as a free poster.
Related Business Pills
All Business Pills
- The Memory Wall: Why a Faster AI Chip Is Not Faster AI
- Tokens per Watt: What an AI Data Centre Actually Produces
- Why AI Can’t Be Both Instant and Cheap: Latency vs Throughput
- The Model and the Harness: Why Same-Model Products Differ
- Context, Not Capability: Why a Smart AI Model Gives Poor Answers
- Discardable Software: When Code Is Cheap Enough to Throw Away
- Human in the Loop vs Human on the Loop: Supervising AI Agents
- The AI Audit Problem: When Making Work Is Cheaper Than Checking It
- AI Evaluation as Acceptance Test: How to Know It’s Good Enough
- Extensibility Is Control: Who Holds the Power in an AI Product
- RLHF Explained: How an AI Model Learns What People Prefer
- Pretraining Explained: How an AI Model Learns Before Anyone Teaches It
- Fine-Tuning Explained: How to Adapt a General AI Model to One Job
- Tokens Explained: The Unit AI Reads, Writes and Bills In
- Embeddings Explained: How a Machine Compares Meaning
- RAG Explained: How a Model Answers From Your Documents
- AI Hallucination Explained: Why Models State False Things
- Distillation Explained: How a Small Model Learns From a Large One
- Reasoning Models Explained: What Changes When AI Thinks First
- Tool Use Explained: How an AI Model Goes From Text to Action
- Capex and Depreciation Explained: Why Chip Lifetime Drives AI Profits
- Run-Rate Revenue Explained: What an AI Company’s Number Means
- Backlog Explained: Revenue That Is Signed but Not Yet Earned
- Switching Costs Explained: Why It Is Hard to Leave an AI Supplier
- Temperature Explained: The Dial That Sets How Predictable an AI Model Is
- Guardrails Explained: Rules Enforced Around an AI Model, Not Inside It
- Benchmarks Explained: Why a Test Score Is Not Your Own Result
- Agent Memory Explained: How an AI Assistant Remembers You Between Chats
- MCP Explained: The Shared Plug That Connects AI Models to Tools
- Quantization Explained: How a Large AI Model Is Shrunk to Fit
- Parameters Explained: Where an AI Model’s Knowledge Lives
- Scaling Laws Explained: Why AI Labs Keep Building Bigger
- Synthetic Data Explained: When a Model Writes Its Own Training Examples
- Transformers and Attention Explained: How AI Reads Every Word at Once









