Adidas Sells European Sportswear on American Cloud.
Germany's flagship sportswear company runs its ERP on AWS, its storefront on Salesforce, and its data on Databricks. Three US clouds, one German brand.
Key Takeaways
- Adidas migrated SAP S/4HANA to AWS in November 2021. The initial load was 9.4 billion financial transactions in one month.
- The data and AI platform runs on Databricks. The storefront runs on Salesforce Commerce Cloud. The edge runs on Akamai.
- Adyen handles payments and is the only European vendor with a load bearing role.
- Score 17 out of 100. Badge: American with European Branding.
Why Adidas Matters
Adidas is German manufacturing in the public imagination. The company was founded in Herzogenaurach in 1949, lists on the DAX, and runs its global operations from the World of Sports campus at Adi Dassler Strasse 1 in the same town it started in. The Arena building, inaugurated in 2019 for the brand's 70th anniversary, is the closest thing global retail has to a Bauhaus monument.
The brand is German. The infrastructure is not.
Adidas runs its global ERP on AWS, its data platform on Databricks, its e commerce storefront on Salesforce Commerce Cloud, and its edge protection on Akamai. The European vendor on the public stack is Adyen, for payments. Everything else is American.
This is part four of our series. Spotify scored 12, Booking.com 40, Klarna 11. Adidas is the most US dependent stack in the series so far.
Methodology. Same as previous parts: eight infrastructure layers scored zero to ten on European ness, weighted by sovereignty impact. As a DAX listed German company, Adidas is the closest thing the public markets have to a national champion. The German federal government has been one of the loudest voices in Europe on digital sovereignty, championing Gaia-X and STACKIT. Adidas's stack, fully documented in public AWS, Databricks, and Salesforce case studies, sits at odds with that national posture.
The Stack, Layer by Layer
Compute and Orchestration
Adidas adopted Kubernetes in 2017 with Giant Swarm as the platform partner. The estate scaled to around 4,000 pods, 200 nodes, and 80,000 builds per month. Within six months of the rollout, 100 percent of e commerce was on Kubernetes. By 2024, Karpenter for autoscaling and Kyverno driven vertical pod autoscalers had reduced compute cost by roughly 50 percent. Clusters run across AWS and on premises. The case study materials lean heavily on AWS. Host: AWS plus on premises. Layer score: 2 of 10.
Storage and Databases
The named system is a cloud data lake on AWS, referenced as part of the SAP S/4HANA modernisation. Specific OLTP database engines are not publicly enumerated. Host: AWS. Layer score: 0 of 10.
Data Warehouse and Analytics
Adidas's data platform runs on the Databricks Lakehouse, with Delta Lake, Unity Catalog, Mosaic AI Vector Search, and Databricks BGE Large embeddings. Adidas open sourced its lakehouse-engine Python and Spark library on GitHub. The Databricks Data plus AI Summit 2025 featured an Adidas talk on agentic workflows. Databricks is US incorporated and runs on AWS, Azure, and GCP. Adidas's deployment is on AWS. Host: AWS through Databricks. Layer score: 0 of 10.
AI and Machine Learning
Adidas runs generative AI on more than 2 million product reviews using Databricks and Mosaic AI Vector Search. ML driven product recommendations and Complete the Look use a partnership with Findmine. Virtual Try On uses computer vision and NeRF based 3D models, with the 3D content pipeline built on NVIDIA RTX in partnership with Covision Media. Hundreds of operational predictive models run on the Lakehouse. Host: Databricks on AWS, NVIDIA for content generation infrastructure. Layer score: 0 of 10.
Observability and Monitoring
Prometheus is named in the Kubernetes case study as part of the cloud native platform. There is no primary public evidence of a SaaS observability vendor at Adidas scale. Host: Prometheus co resident with Kubernetes clusters on AWS and on premises. Layer score: 3 of 10.
Identity and Access
Customer identity, including SSO, social login and MFA, is referenced in Adidas digital programme materials but the specific vendor is not publicly disclosed in primary sources. Workforce identity provider is not publicly disclosed. Layer score: not publicly evidenced.
CDN and DNS
Akamai is the publicly confirmed CDN, including Akamai Content Protector for bot mitigation. The 2025 Akamai case study confirms Adidas Switzerland specifically as a customer. Akamai is headquartered in Cambridge, Massachusetts. Host: Akamai global edge. Layer score: 0 of 10.
Payments
Adyen is the publicly confirmed processor, announced as an Adyen enterprise client in October 2018. Host: Adyen (Amsterdam, EU regulated). Layer score: 9 of 10.
ERP and E commerce (additional layers)
Adidas migrated its SAP environment from ECC 6.0 to SAP S/4HANA on AWS, announced in November 2021. The initial load was 9.4 billion financial transactions in one month. The e commerce storefront runs on Salesforce Commerce Cloud (formerly Demandware), an Adidas customer since 2011, with Marketing Cloud and Service Cloud (1,000 plus care agents) sitting alongside.
Sovereignty Scorecard
| Layer | Weight | Score |
|---|---|---|
| Compute and orchestration | 20 | 2 |
| Storage and databases | 18 | 0 |
| Data warehouse and analytics | 14 | 0 |
| AI and machine learning | 12 | 0 |
| Observability and monitoring | 8 | 3 |
| Identity and access | not scored | unverified |
| CDN and DNS | 10 | 0 |
| Payments | 10 | 9 |
Weighted total across the scored layers: 17 out of 100.
Badge: American with European Branding.
The Kubernetes plus Prometheus base offers a degree of portability that several other companies in this series do not have. But the substrate is AWS, the data platform is Databricks, the storefront is Salesforce, and the edge is Akamai. There is no plausible reading of this stack that calls it European in any meaningful sense.
What a Fully European Stack Would Look Like
- Compute and orchestration. Managed Kubernetes from Scaleway, OVHcloud, STACKIT, or IONOS Cloud. STACKIT in particular is operated by the Schwarz Group, parent of Lidl, and was built specifically for European enterprise workloads with German data centre presence.
- Storage and databases. Managed Postgres, MySQL, and Kafka from Aiven (Helsinki). Object storage from any EU hyperscaler. SAP S/4HANA can be deployed on STACKIT, OVHcloud, or T-Systems sovereign cloud as drop in replacements for AWS hosted SAP.
- Data warehouse and analytics. ClickHouse on European infrastructure, or self managed Apache Spark with Delta Lake on object storage, gets functionally close to what Databricks provides. The Adidas open source lakehouse-engine is already portable.
- AI and ML. Mistral AI (Paris) for general purpose LLM workloads. Aleph Alpha (Heidelberg) for enterprise and regulated deployments. NVIDIA hardware can be procured in European data centres via OVHcloud GPU or Scaleway AI.
- Observability. Self hosted Grafana and Prometheus.
- CDN. BunnyCDN (Slovenia) or CDN77 (Czech Republic) replace Akamai for content delivery. Bot mitigation is a harder gap; the European equivalents are thinner.
- Payments. Already Adyen.
- E commerce. Shopware (Germany) and PrestaShop (France) are the open source European peers. Saleor (Poland) is API first and headless.
See the cloud alternatives catalog, the Alternatives to AWS page, and the e commerce platforms hub.
Why This Is Hard
Adidas's digital transformation under Kasper Rorsted (2016 to 2022) was an explicit strategic priority. The company targeted e commerce growth from one billion euros in 2016 to four billion euros by 2020, and hit it. The chosen tools were AWS, Salesforce, and Databricks because in 2017 and 2018 those were the tools that demonstrably moved fastest. The European alternatives were either smaller, less feature complete, or did not yet exist.
The SAP S/4HANA migration in 2021 reinforced the lock in. Moving 9.4 billion financial transactions onto a new ERP is the kind of project that defines a CIO career. Doing it twice in five years to switch from AWS to STACKIT is not a project most boards will approve.
The strongest counterargument inside Germany is the Schwarz Group itself, parent of Lidl, which built STACKIT precisely because it refused to make Adidas's choice. The capability gap that existed in 2017 is smaller in 2026. Whether Adidas revisits the decision in the next ERP cycle is the more interesting question.
We did not find any public Adidas statement on the US CLOUD Act, EU data residency, or sovereign cloud. See our coverage of the US CLOUD Act and the broader digital sovereignty debate.
Frequently Asked Questions
Is Adidas data subject to the US CLOUD Act?
Adidas is German, but AWS, Databricks, Salesforce, and Akamai are all US companies. The CLOUD Act applies to data held by US companies regardless of physical location. Adidas main mitigation is contractual (EU data residency commitments from the vendors) rather than structural.
Does Adidas participate in Gaia-X?
No primary public evidence places Adidas on Gaia-X member or working group rosters. The German government sovereign cloud push has not changed Adidas published vendor mix.
Why does Adidas use Databricks instead of a European alternative?
The public material does not address the choice directly. Databricks is the market leader for Lakehouse architectures and was so in 2018 when Adidas data programme was being built. Aiven for ClickHouse and self managed Apache Spark with Delta Lake are the closest European equivalents in 2026.
Could Adidas move to STACKIT?
STACKIT was built for exactly this kind of customer. The technical lift is large but not unprecedented; the political tailwind is real, especially with the Schwarz Group as a peer DAX adjacent German organisation championing the alternative.