THE SEEN
200M+
Protein structure predictions
The AlphaFold database made predictions available at scientific scale. This is a research count, not sales. [1]
The research people can use freely. The paid tools built around it.
Google DeepMind develops AI models. Some results are shared freely; commercial use can happen through other offers within Google. A free research database, paid model processing and a commercial AlphaFold subscription serve different needs. Follow the actual offer and the organisation that sells it.
Research and Google commercial offers checked 9 November 2026. This is a Google DeepMind technology map within Google, not a claim that all Google Cloud revenue belongs to a standalone DeepMind business.
THE SEVEN MONEY GATES CANVAS
2 Money Gates in UseOne brand. Physical and digital.What do you think each payment buys?
Explore the exchanges ↗| MONEY GATE | PHYSICAL | DIGITAL |
|---|---|---|
| 1Products | — | — |
| 2Services | — | |
| 3Access | — | |
| 4Attention | — | — |
| 5Money / Float | — | — |
| 6Risk | — | — |
| 7Brand / IP | — | — |
THE SEEN
200M+
The AlphaFold database made predictions available at scientific scale. This is a research count, not sales. [1]
THE UNSEEN
Commercial access
Google Cloud documents a commercial subscription and allowlist process. No public DeepMind revenue total is inferred. [2]
Figures retain their stated period and scope. The gate map interprets the documented offers; company totals are not separate gate accounts.
THE EXCHANGE
A researcher can use freely shared AlphaFold results. A commercial organisation seeking AlphaFold 3 access follows a different route through Google Cloud.
Google also sells processing with Gemini models. The customer pays for useful work performed by the model, under the provider’s service terms.
THE EVIDENCE
AlphaFold’s database contains more than 200 million predictions. That scale demonstrates reach and usefulness; it does not tell us what a commercial customer pays.
The Cloud documentation makes the paid route concrete: a commercial subscription and an allowlist. Gemini pricing separately describes model-processing charges.
THE DECISION
A useful invention can become a service, a paid use right or a licence. Those are choices about what the customer receives.
Here, the documented subscription and processing offers are the examples. Isomorphic Labs collaborations and Google Cloud sales should keep their own provider and contract identities.
Think about the question. Then reveal what is sold, who pays and why.
Sell a physical or digital item.
Get paid to do useful work.
THE EXCHANGE
Customers pay Google Cloud to process requests using Gemini models developed by Google DeepMind.
THE EVIDENCE
Google Cloud publishes processing prices that vary by model and input/output type. [3]
THE DECISION
Useful model work can be sold by measured usage. The billing unit does not make it a sale of the customer’s own permanent model.
Sell the right to use something.
THE EXCHANGE
An eligible organisation obtains a commercial subscription through Google Cloud.
THE EVIDENCE
The documentation requires a commercial subscription and allowlisting. Non-commercial academic access follows different terms. [2]
THE DECISION
Separate public research access from a commercial right to use the capability. Name the provider and the permitted use.
Get paid to reach an audience.
Earn through capital and payment flows.
Get paid to cover defined uncertainty.
Earn from licensed identity or know-how.
Physical and digital are two forms of the same gate. Each gate is counted once. A blank box means this dossier has not added an example in that form. The gate readings are Pelumi’s application of The Seven Money Gates, not the companies’ accounting labels.
What the customer buys. What else the brand earns from. What you can learn.
A researcher may need shared predictions. A business may need commercial access or model processing.
Google’s documented paid subscriptions and processing offers around DeepMind technology.
Keep your open resource useful. Design a distinct commercial offer around a customer need you can deliver.
Research timeline, 200M+ protein predictions and free non-commercial research server.
Commercial access requires a subscription and allowlisting through the Google Cloud account team.
Paid Gemini processing through Google Cloud; model, modality and service terms affect the charge.