---
title: "What is a subprocessor? GDPR Article 28 and LLM APIs"
description: "A subprocessor is another company your processor uses to handle your personal data. What GDPR Article 28 requires and how it applies to LLM API GPU hosts."
url: https://broadrouter.com/glossary/subprocessor/
lang: en
updated: 2026-11-15
publisher: BroadRouter (a BroadNet site); API by BroadNet Inference, https://inference.broadnet.ai
---

# Subprocessor

Updated 2026-11-15

A subprocessor is a third party that a processor engages to carry out processing of personal data on behalf of the controller. For an LLM API, subprocessors can include GPU hosts, partner inference providers and backup capacity. Under GDPR Article 28 they need the controller's authorisation and the same data protection obligations.

Also called: sub-processor

Your contract is with one provider. Your prompt may still touch three companies. Subprocessors are the ones you do not see.

## How it works

The GDPR sets up a chain. You are usually the controller, the LLM provider is your processor, and anyone the provider uses to handle your data is a subprocessor (the regulation says “another processor”). Article 28 sets two rules:

-   **Authorisation (28(2)).** “The processor shall not engage another processor without prior specific or general written authorisation of the controller.” With general authorisation, the processor must inform you of intended additions or replacements, “thereby giving the controller the opportunity to object”.
-   **Same obligations (28(4)).** The data protection obligations in your contract must be imposed on the subprocessor by contract. If the subprocessor fails, “the initial processor shall remain fully liable to the controller”.

In practice, providers publish a list and a way to follow changes. Mistral’s help center says its list “is available at all times on our Trust Center” and that customers can subscribe to email notices when a subprocessor is added or removed. Its DPA gives general authorisation and a ten-day window to object in writing.

A typical LLM API request path might look like this:

| Step | Who | Subprocessor? |
| --- | --- | --- |
| Your app | You | No, you are the controller |
| API gateway | The provider | No, the processor |
| GPU host or partner model | Cloud or inference partner | Yes, if it processes the prompt |
| Failover capacity | Another host | Yes |

## Why it matters

-   **Location.** Each subprocessor can be in a different country, which matters for [data residency](https://broadrouter.com/glossary/data-residency/) and transfer rules.
-   **Retention.** A provider’s [zero data retention](https://broadrouter.com/glossary/zero-data-retention/) promise only holds if each subprocessor follows it.
-   **Routers add layers.** A hosted [LLM router](https://broadrouter.com/glossary/llm-router/) sits between you and the model provider, so check both.

The [GDPR checklist](https://broadrouter.com/blog/gdpr-llm-api-checklist/) lists the questions to ask: who the subprocessors are, including GPU hosts and backup capacity, and how you are told about changes.

## On BroadNet Inference

BroadNet’s Terms (§7.3a) say: “Partner-hosted and backup processing involves other processors. A current subprocessor list is available to customers on request.” The privacy statement says the same. BroadNet does not publish the list on a public page, so ask for it through the [contact page](https://broadrouter.com/contact/) before you send regulated data.

What BroadNet does state publicly: its own and hosted open-weight models run on BroadNet-owned or leased hardware in the United States, with partner capacity as backup. Partner-served models, such as [DeepSeek V4.1 Flash](https://broadrouter.com/models/deepseek-v4.1-flash/), run under zero-data-retention terms with US inference partners. This is a summary, not legal advice.

## Questions

### Is a cloud GPU host a subprocessor of an LLM API?

Usually, if it processes your prompts and those prompts contain personal data. Ask the provider for the full list, including partner inference providers and backup capacity, not only its main cloud.

### Can I object to a new subprocessor?

Under GDPR Article 28(2), where you give general written authorisation the processor must tell you about intended changes and give you the opportunity to object. The notice period and process are set in the DPA; Mistral's DPA, for example, allows a written objection within ten days of notice.

### Does BroadNet publish a subprocessor list?

Not publicly. Its Terms and privacy statement say a current subprocessor list is available to customers on request.

## See also

-   [Data processing agreement (DPA)](https://broadrouter.com/glossary/dpa/) (term)
-   [GDPR](https://broadrouter.com/glossary/gdpr/) (term)
-   [Data residency](https://broadrouter.com/glossary/data-residency/) (term)
-   [Zero data retention](https://broadrouter.com/glossary/zero-data-retention/) (term)
-   [Inference provider](https://broadrouter.com/glossary/inference-provider/) (term)
-   [LLM router](https://broadrouter.com/glossary/llm-router/) (term)
-   [GDPR and LLM APIs: 15 questions to ask your provider](https://broadrouter.com/blog/gdpr-llm-api-checklist/) (article)
-   [Zero data retention LLM API: a 10-question checklist](https://broadrouter.com/blog/zero-data-retention-llm-api-checklist/) (article)
-   [A private endpoint starts with a clear workload](https://broadrouter.com/blog/private-endpoint-planning/) (article)

Try it on your real workload. New accounts start with $5 of credit for Gemma and transcription. DeepSeek, GLM and Qwen3.8 need a top-up.

[Get started](https://inference.broadnet.ai/account/?utm_source=broadrouter&utm_medium=referral&utm_campaign=glossary-subprocessor&utm_content=page-cta) ·

**Sources**

-   [Regulation (EU) 2016/679 (GDPR), Article 28(2) and 28(4), official text (checked 2026-11-15)](https://eur-lex.europa.eu/eli/reg/2016/679/oj)
-   [Mistral AI Help Center, who are your subprocessors and how can I stay updated (checked 2026-11-15)](https://help.mistral.ai/en/articles/455208-who-are-your-subprocessors-and-how-can-i-stay-updated-on-changes)
-   [Mistral AI Data Processing Addendum, subprocessor authorisation and objection (checked 2026-11-15)](https://legal.mistral.ai/terms/data-processing-addendum)
-   [BroadNet Terms of Service, §7.3a hosting and processing locations (checked 2026-11-15)](https://www.broadnet.ai/terms)
-   [BroadNet Inference privacy statement (effective September 2026; checked 2026-11-15)](https://www.broadnet.ai/inference-privacy)
