Ethical Stack

Cerebras Inference

Inference-as-a-service runs models on Cerebras infrastructure; model provider terms may govern output and model-specific rights.

Score calculation

(7 + 4) ÷ 2 = 5.5

2 of six dimensions rated. Unrated dimensions and BDS status are excluded.

Full scoring method ↗
Setup details

Cerebras is hardware/infrastructure and inference service operator; specific hosted model is third-party in some API uses.

Energy

↑

Cerebras announced a Finland data center scaling to 165 MW, with closed-loop cooling and planned heat recovery; this is infrastructure capacity and design information, not actual electricity/water use per Cerebras Inference request.

Reasoning & sources

Cerebras disclosed a substantial Finland capacity plan and design measures for water reuse and heat recovery, but not metered operations or API-level allocation. Request the selected endpoint's region, utilization, energy, and water reporting; do not attribute all facility capacity or promised efficiency to a particular inference request.

The company announcement is forward-looking; it does not disclose operational utilization, regional grid emissions, water data, or model/request allocation.

Training rights

↑

Cerebras’ terms defer API output ownership to third-party model terms and disclaim Cerebras ownership; they do not establish source-data rights for underlying models.

Reasoning & sources

Cerebras’ terms defer API output ownership to third-party model terms and disclaim Cerebras ownership; they do not establish source-data rights for underlying models. Select the model first and review its license plus corpus-provenance disclosures; prompt no-training terms do not clear the model’s original dataset.

Identify each model developer/license and investigate corpus provenance independently.

Privacy

↑

Cerebras’ privacy notice states inputs/outputs for its training, inference and chatbot services are not retained as described in that policy; API third-party model terms also apply.

Reasoning & sources

The company says inputs/outputs are not retained under its policy, but downstream third-party model terms still apply. Check the exact API/web surface, plan, region, retention setting, routing, and stateful features in the applicable terms before sending sensitive content.

The company says inputs/outputs are not retained under its policy, but downstream third-party model terms still apply.

Exact API/model, retention exceptions, processor/subprocessor terms and region need review.

Openness

↑

Cerebras offers hosted inference for selected models and an enterprise route for custom weights; its serving stack is not shown as portable.

Reasoning & sources

The catalog includes open models, but the API runtime is managed and portability is limited. Check whether the chosen model’s weights and license permit inspection or relocation, and whether the serving runtime moves too; API compatibility alone is not an open stack.

The catalog includes open models, but the API runtime is managed and portability is limited.

The product page does not establish open-source Cerebras serving software, model training-data provenance, or availability of custom-weight deployment to all customers.

Worker treatment

↑

Cerebras' Business Partner Code sets expectations on wages, working hours, safety, forced labor, and human rights for suppliers and contractors. It requires records but the reviewed source does not report labor audits, worker outcomes, or model-data annotation conditions.

Reasoning & sources

Cerebras has a written business-partner code covering labor standards, but this is a commitment and recordkeeping requirement rather than a public audit result. Ask for recent supplier assessments, remediation closure data, and annotation-worker pay/location/exposure information before treating the code as proof of better working conditions.

No supplier audit findings, remediation outcomes, annotation vendor locations, or worker-pay data are disclosed in the cited code.

Government & military

↑

Cerebras says it joined the Department of Defense's Tradewinds Solutions Marketplace and partnered with Carahsoft to make its AI infrastructure available to federal agencies and defense organizations. The announcement establishes a procurement route, not an awarded deployment, named program, or specific inference workload.

Reasoning & sources

Cerebras documents entry into a DoD procurement marketplace and a government reseller channel, but does not identify a specific award or operational use of Cerebras Inference. Request the award, customer, endpoint, and use restrictions; marketplace eligibility should not be presented as deployment evidence.

The public announcement does not identify a military contract award, actual customer, operational endpoint, or safeguards review.

BDS status

↑

We did not find Cerebras Systems, Inc. among the named targets in the two official BNC campaign lists. This is a limited list check, not BDS approval.

Reasoning & sources

Cerebras Systems, Inc. is not a named target in the checked lists, but BDS uses selective campaigns, so absence cannot establish a clean record. Your selected model, cloud and billing route can introduce other companies into the purchase.

This review covers the named targets in the BNC consumer/pressure guide and No Tech campaign, not every BDS campaign, local group, investment or supplier. Lists and ownership can change. A company’s absence from these lists is not an endorsement or proof that it has no relevant relationships.

All sources

Side by side