DataRobot licenses an AI platform that its legal documents describe as a standardised commercial B2B software platform on which customers create their own predictive models using various DataRobot product components.[1] The same legal documents now cover generative AI products, with third-party large language models and vector databases offered as Third Party Products on certain Orders.[2] DataRobot publishes its standard contract documents, which makes the contract side of licensing unusually visible, but it does not publish a price list or part numbers, and the quantities that matter (seats, workers, active users, deployments) are set by each Order and licence key.
For a software asset manager the first question is which agreement governs. A customer with an Order is under the Master Subscription Agreement (MSA).[2] A self-service or click-through user of the hosted platform is under the AI Platform Agreement.[3] Trial users are under the Evaluation Terms,[4] and anyone using a no-fee tool is under the Tool and Utility Agreement.[5] DataRobot’s legal index also lists a Support Policy, an Availability Policy, a Data Processing Policy, an Information Security Policy, an AI Policy, a reseller and referral partner agreement and Japan-specific versions.[6]
Editions
DataRobot does not publish a named edition ladder in the documents reviewed. The platform is described by delivery model rather than by edition. The MSA defines SaaS as software hosted in and delivered from DataRobot’s managed cloud environment, and on-premise as a version deployed in the customer’s data center or in the customer’s virtual private cloud.[2] The architecture documentation lists the deployment options as multi-tenant SaaS on AWS within three regions, single-tenant SaaS on AWS, GCP and Azure in many regions, self-managed installations in VPCs on AWS, GCP and Azure, and self-managed installations in a wide variety of data centers.[7] The cover note says the MSA works for both on-premise and SaaS deployments if the customer switches at any time.[1]
Within a subscription, capabilities are gated by the licence key. The licence page of a self-managed cluster lists the features and functionality included in the subscription after the key is validated,[8] and the documentation refers customers to DataRobot Support for the meaning of licence details because they are based on the subscription.[9] Seat licences are described as a premium feature that needs a seat allocation in the subscription.[10] See Self-Managed AI Platform, Single-tenant SaaS and Multi-tenant SaaS.
Metrics
The contract defines Authorized Users as the employees, agents and independent contractors of the Customer and of its Affiliates, and the Customer may permit them to use the Solution for the same purposes it may.[2] Numbers appear in the product. The licence page shows the expiration date, concurrent workers limit, maximum active users, prepaid deployment limit, maximum deployment limit and in some cases subscription features.[9] The main measures are:
- Seats and active users. Seat licences grant access to specific functionality and are allocated to organizations and assigned to users; non-builder users do not count towards the maximum active user allocation.[10]
- Workers. Workers are the processing power for projects, model training and predictions. Each user has four workers by default, and an organization can cap the workers its members share.[11]
- Deployments. For MLOps, the deployments limit is set per organization and cannot exceed the prepaid deployments; inactive deployments do not count.[12][13]
- Compute. The maximum compute instances for a serverless platform default to 8 per organization when no limit is set.[12]
- Core usage and LLM usage. A self-managed Usage Report charts CPU and GPU use against the organization’s licence limit for core usage, and the Usage explorer reports LLM API use by service.[14][15]
- Credits. The AI Platform Agreement ties hosted use to unexpired Credits that are payable in advance and non-refundable.[3]
Each measure is described in DataRobot seats, workers and deployment limits.
Counting and floors
No public document sets a minimum. The MSA says order details, including how entitlements are measured and the Subscription Term, are as stated in the Order, and where a customer buys through an authorized partner they are as stated in the partner’s order communicated to DataRobot.[2] The MSA also provides that if it conflicts with an Order, the Order controls for the Solution and Professional Services it covers.[2] Pre-purchased Professional Services and expenses that are not used expire 12 months after purchase unless agreed in writing.[2]
Fees for Maintenance and Support are included in the fees for the Solution, and the Solution, Professional Services and tax are invoiced annually in advance unless the Order says otherwise, with payment within 30 days of the invoice date.[2] Following no less than 14 days’ written notice, DataRobot may suspend access if undisputed payments are not received within 60 days of the invoice date.[2]
Virtualization and partitioning
The MSA does not state a virtualization or partitioning rule. For on-premise use the licence key sets limits on concurrent workers, active users and deployments, and the administrator Usage Report compares CPU and GPU consumption with a licence limit for core usage.[9][14] The documentation says an organization’s worker allocation is independent of the cluster configuration and warns administrators not to oversubscribe the cluster’s physical capacity.[12] A deployment in a customer’s virtual private cloud is on-premise under the MSA.[2]
Cloud and BYOL
Hosted use is under the MSA when the customer has an Order, and the Availability Policy applies only to the SaaS version, both single-tenant and multi-tenant.[16] For SaaS the MSA adds data restrictions: the customer must not import payment card data, financial account credentials, government ID numbers, sensitive personal data or personal data of individuals under 16, and HIPAA-regulated information is allowed only when the Order expressly permits it under the referenced Business Associate Agreement.[2] DataRobot may delete offending data or suspend access.[2] The AI Platform Agreement for self-serve hosted use adds that the customer grants DataRobot a non-exclusive licence to use Customer Data to improve and enhance the Services.[3] The two agreements therefore differ on data rights, which matters when a team starts on a self-serve account and later signs an Order. See DataRobot SaaS and self-managed deployments.
For generative AI, the MSA says that where the Order includes a DataRobot generative AI product the customer may be granted limited access to Third Party Products for evaluation; DataRobot gives no warranty for them and they may change during the Subscription Term.[2] The AI Policy adds that users of those products are also subject to the provider’s own terms as though using the product directly.[17]
Programs
- Fixed term. DataRobot is licensed for a fixed term and purchases cannot be terminated for convenience; Orders are non-cancellable and fees non-refundable except as the MSA provides.[1][2]
- Support and Maintenance. Support is requested through the Support Portal. The initial response target is within 1 hour for Severity 1, 2 Business Hours for Severity 2, 8 Business Hours for Severity 3 and 2 Business Days for Severity 4.[18]
- Availability. For SaaS, the Availability Policy sets a quarterly uptime of at least 99.95% with service credits of 5, 10 or 20 percent of the quarter’s fees for the affected Solution depending on the band.[16] The MSA cover note describes the guarantee as 99.9%, so the policy document should be treated as the operative text.[1]
- Evaluation. Evaluation Software is for 30 days of internal evaluation, with models deleted unless the Solution is purchased within three months.[2]
Audits and compliance
The MSA reviewed has no licence audit or verification clause. Compliance evidence sits in the product: the licence page, the seat allocation pages and the usage reports.[9][10][14] The restrictions in the MSA are, however, a ground for immediate termination in part: DataRobot may terminate immediately upon breach of the reverse-engineering restriction or of the customer data restrictions in s.12.5 or s.12.6(a).[2] Where a self-managed licence expires before renewal, users can continue to make predictions with existing models but cannot build new models or insights or generate Compliance Documentation.[9]
Out of scope
This article does not cover DataRobot’s open source projects, the pricing of any deal, the Public Sector Services Terms or Services Packages documents listed on the legal page, or the terms of third-party model providers beyond noting that they apply.[6]