Hikube by Hidora vs Amazon Web Services (AWS)

The Swiss Kubernetes-First Sovereign Cloud vs America’s Default

Hidora SA — registered in Geneva under CHE-286.910.173 — has been running Kubernetes as a service in Switzerland since 2016. Founded and run by CEO Matthieu Robin, with a specific engineering focus on Kubernetes-native infrastructure. It is a KCSP (Kubernetes Certified Service Provider — verifiable in the CNCF landscape) and a CNCF Silver member — signals that Kubernetes is not marketing here, it is the core competency.

Hikube is Hidora’s complete sovereign Swiss cloud: managed Kubernetes, cloud instances, GPU-as-a-Service, S3-compatible object storage, managed databases, and streaming/messaging — running exclusively in three Swiss data centres (Geneva, Gland, Lucerne) with native synchronous replication across all three. The stack is 100% open source: Kubernetes on Talos, KubeVirt for the VM layer, SeaweedFS for object storage. No proprietary APIs to build against; no vendor-specific control plane; a workload built on Hikube is a workload that runs on any Kubernetes.

Amazon Web Services is a business unit of Amazon.com, Inc. — Delaware-incorporated, Nasdaq-listed (AMZN), headquartered in Seattle, Washington. AWS is the market-default cloud with an unmatched service catalogue depth. Its EU regions provide data residency; its AWS European Sovereign Cloud (now live, with its first region in Brandenburg, Germany) targets regulated European workloads. Neither removes the underlying reality that Amazon is a US parent, and CLOUD Act plus FISA Section 702 reach applies to customer data it holds anywhere.

For Swiss and European organisations where the sovereignty question is about who legally holds the compute and who legally controls the stack — Hikube’s Swiss parent, Swiss data centres, and open-source stack answer both structurally.

Jurisdiction Comparison

	Hikube by Hidora	Amazon Web Services (AWS)
HQ	Geneva, Switzerland	Seattle, Washington, USA
Corporate entity	Hidora SA (Swiss commercial register — CHE-286.910.173)	Amazon.com, Inc. (Delaware; Nasdaq: AMZN)
Founded	2016 (Matthieu Robin, CEO)	AWS 2006; Amazon.com 1994
CLOUD Act exposure	None (Swiss jurisdiction, no US parent)	Yes (US-incorporated parent)
FISA 702 exposure	None	Yes
Data centres	3 Swiss DCs (Geneva, Gland, Lucerne), native synchronous replication	30+ global regions (EU + AWS European Sovereign Cloud, first region live in Brandenburg, Germany)
Stack	100% open source (Kubernetes/Talos, KubeVirt, SeaweedFS)	Proprietary (S3, DynamoDB, Lambda, EventBridge, etc.)
Egress fees	Free (both inbound + outbound)	Per-GB metered, destination-dependent
Exit fees	None	None (data transfer costs apply)
Certifications	ISO 27001 (SQS), Swiss Hosting, Swiss Label, KCSP, CNCF Silver	ISO 27001, SOC 1/2/3, FedRAMP, HIPAA, PCI DSS
Sovereign certification	Swiss Hosting (direct); Swiss jurisdiction by construction	AWS European Sovereign Cloud (live; first region in Brandenburg, Germany)

Product Capabilities

Managed Kubernetes

• Hikube: KCSP-certified Managed Kubernetes on Talos, integrated with cross-DC S3 storage, managed databases, and GPU nodes. Standard tooling (Helm, kubectl, ArgoCD, Terraform). Control plane 32 CHF/month.
• AWS: Elastic Kubernetes Service (EKS) with EKS Anywhere, Fargate serverless nodes, deep IAM Identity Center integration, EKS-optimised AMIs, Bottlerocket OS.

Notable: EKS has more managed features and deeper IAM integration; Hikube’s KCSP-grade Kubernetes is production-ready with a fully open-source substrate and predictable flat-rate control-plane pricing.

GPU-as-a-Service — Published Price List

• Hikube: Full public GPU price list — NVIDIA L4 at 165 CHF/month, L40S at 760 CHF/month, A100-80 at 1,495 CHF/month, RTX 6000 Pro at 1,550 CHF/month, H100 at 1,999 CHF/month, H200 at 2,499 CHF/month (on demand). The current top tier for AI training and inference workloads.
• AWS: P5 instances (H100), P4d (A100), G6 (L4), Inferentia/Trainium (custom silicon), Bedrock (managed model catalogue), SageMaker.

Notable: For teams whose workload requires the same NVIDIA GPU tier, both providers offer H100/H200/A100 access. AWS adds custom-silicon options (Trainium/Inferentia) and a managed model catalogue (Bedrock) that Hikube does not directly replicate. For open-weight model training and inference in Swiss jurisdiction with fully-published monthly pricing, Hikube is the direct alternative. Hikube’s price transparency contrasts sharply with AWS GPU pricing where Reserved Instances, Savings Plans, and Spot combine into a complex negotiation.

Object Storage

• Hikube: S3-compatible via SeaweedFS backend, with cross-DC replication across Swiss data centres included. 0.02 CHF/GB/month standard, 0.03 CHF/GB/month encrypted. Existing S3 tooling migrates without code changes.
• AWS: S3 (proprietary API but the industry reference for S3 compatibility), 11 nines durability, deep tiering (Standard, IA, Glacier, Deep Archive). Egress metered.

Notable: S3 is the industry reference; SeaweedFS’s S3 compatibility is a strong subset. For applications using S3-compatible tooling (rclone, s3cmd, AWS CLI, Boto3, Terraform), migration is straightforward. Hikube pricing includes cross-DC replication in the base rate.

Managed Databases

• Hikube: Managed PostgreSQL, MySQL, Redis, Kafka, RabbitMQ from 4 CHF/month (nano tier) up to 160 CHF/month (2xlarge). Standard open-source engines with no proprietary API layer.
• AWS: RDS (PostgreSQL/MySQL/SQL Server/MariaDB/Oracle/Aurora), DynamoDB (proprietary NoSQL), ElastiCache (Redis/Memcached), MSK (managed Kafka), Amazon MQ.

Notable: Hikube’s DB portfolio covers the standard OSS engines with no proprietary equivalent required — critical for teams that reject vendor lock-in on the data layer. AWS adds proprietary services (DynamoDB, Aurora) that lock workloads to AWS if used natively.

Compute

• Hikube: Cloud Instances via KubeVirt with a 1:4 vCPU/RAM ratio — u1.medium (1 vCPU / 4 GB) at 40 CHF/month, scaling linearly through u1.8xlarge (32 vCPU / 128 GB) at 1,280 CHF/month. GPU nodes and standard Kubernetes worker nodes with autoscaling.
• AWS: EC2 with hundreds of instance types, Fargate (serverless containers), Lambda (serverless functions), Batch, Lightsail.

Notable: AWS’s compute breadth and instance-family depth is unmatched. For Kubernetes-native workloads with predictable pricing, Hikube’s flat monthly rates avoid the AWS complexity. For serverless-heavy architectures, AWS Lambda remains architecturally unique.

Observability + IaC

Both support the standard OSS observability stack (Prometheus, Grafana). Hidora publishes Terraform, Helm, and ArgoCD as first-class tooling. AWS supports Terraform via the AWS provider and CDK for programmatic infrastructure, plus proprietary CloudFormation.

Egress Economics: The Silent Bill

The single largest total-cost-of-ownership difference between Hikube and AWS is egress. AWS meters per-GB outbound traffic by destination region; for egress-heavy workloads — video delivery, downloads, API-heavy backends, ML inference outputs, data pipelines — egress can dominate the monthly bill. Some organisations report egress accounting for 20-50% of AWS spend on egress-heavy workloads.

Hikube publishes egress as free at every tier — both inbound and outbound — with no per-GB metering and no exit fees. For a workload serving even a few TB/month of outbound traffic, the total-cost difference at scale can be significant, and for high-egress workloads (video, downloads, AI inference APIs), the multiplier is dramatic.

Sovereignty Model

The structural distinction is not “EU region” — it is corporate parent and control-plane ownership:

• Hikube: Swiss parent (Hidora SA, CHE-286.910.173); Swiss data centres exclusively; Swiss federal jurisdiction; nFADP compliance; Swiss Hosting certification; FINMA-relevant for finance; no CLOUD Act reach exists as a matter of jurisdictional law.
• AWS: US parent (Amazon.com, Delaware); EU regions and AWS European Sovereign Cloud (live since its first region in Germany) provide data-residency commitments; the CLOUD Act and FISA Section 702 still apply to Amazon as a US corporation regardless of region.

For Swiss banks, insurance, healthcare providers, defence contractors, and public administrations subject to FINMA, HDS-equivalent, or defence-classification frameworks, the corporate-parent distinction is a formal risk-register item — not a marketing preference.

Pricing Snapshot

Workload	Hikube (published, CHF)	AWS (list, USD equivalent)
Cloud instance (1 vCPU / 4 GB)	40 CHF/month	t3.medium ~ $30/month + egress
Cloud instance (8 vCPU / 32 GB)	320 CHF/month	m5.2xlarge ~ $280/month + egress
Managed Kubernetes control plane	32 CHF/month	EKS ~ $73/month
Managed PostgreSQL nano	4 CHF/month	RDS db.t3.micro ~ $15/month
S3-compatible storage (per TB)	20 CHF/TB/month	S3 ~ $23/TB + egress
Egress (per TB)	Free	~$80-90/TB (destination-dependent)
GPU H100	1,999 CHF/month (1× H100)	p5.48xlarge bundles 8× H100, so not like-for-like; AWS cut P5 on-demand prices by up to 45% from June 2025 — check the current EC2 price list
GPU L40S	760 CHF/month	G6 varies (~$1-2/hour)

For like-for-like workloads Hikube is materially cheaper on published rates; on egress-heavy workloads the gap widens dramatically because Hikube egress is free while AWS egress is metered per-GB. GPU pricing model differences are the biggest gap — Hikube publishes flat monthly rates; AWS uses hourly with heavy dependence on Reserved Instance / Savings Plan discounts.

Reference Customers

Public Hidora reference customers include DATEA Services, Novel-T, ECAB, Adopt-AI, DeepLink, Viteos, Sisteo, and Alpes-Habitat — spanning Swiss engineering, health-tech, insurance, and municipal services.

Who Should Switch?

Hikube is the structural choice for:

• Swiss banks and insurance companies subject to FINMA requirements where cloud vendor jurisdiction is auditable
• Swiss healthcare providers, hospitals, and medical research organisations where patient-data residency in Switzerland is a legal requirement
• Swiss public administrations with sovereign-cloud procurement policies
• AI teams training or serving models on open-weight architectures who want H100/H200/A100/RTX 6000 Pro in Swiss jurisdiction, with predictable monthly pricing
• Kubernetes-first engineering teams who value KCSP-grade engineering depth and a fully open-source stack
• European organisations subject to NIS2, DORA, or CRITICAL-INFRASTRUCTURE frameworks that document cloud-vendor jurisdiction
• Egress-heavy workloads — video streaming, downloads, AI inference APIs, data-pipeline outputs
• Reversibility-committed organisations — the 100% open-source stack means workloads remain portable to any Kubernetes environment

AWS remains the right choice for:

• Organisations deeply committed to AWS-proprietary services (DynamoDB, Lambda@Edge, Step Functions, SageMaker, Bedrock) with no re-platforming path
• Global consumer applications requiring the AWS regional and edge (CloudFront, Lambda@Edge) footprint
• Organisations already on multi-year AWS Enterprise Discount Program commitments
• Teams standardised on AWS-native IAM Identity Center and Organizations management

The Bottom Line

Amazon Web Services is the market-default cloud with the deepest service catalogue in the industry. Its EU regions and now-live AWS European Sovereign Cloud offer meaningful data-residency and partial-sovereignty commitments.

None of this changes the underlying jurisdictional reality: Amazon.com, Inc. is a Delaware US corporation, and CLOUD Act plus FISA Section 702 reach applies to customer data it holds worldwide, including EU and Swiss regions.

For Swiss and European organisations where the sovereignty question is about corporate parentage and control-plane ownership — not just region selection — Hikube is the structurally aligned choice. Hidora SA (Geneva, CHE-286.910.173) under Swiss federal jurisdiction; three Swiss data centres (Geneva, Gland, Lucerne) with synchronous replication; 100% open-source stack (Kubernetes/Talos, KubeVirt, SeaweedFS); KCSP + CNCF Silver + ISO 27001 + Swiss Hosting + Swiss Label, plus a FINMA-oriented finance offering; GPU-as-a-Service with L4/L40S/A100/RTX 6000 Pro/H100/H200 at published monthly rates; free egress at every tier.

For organisations where jurisdiction, egress economics, and Kubernetes-native engineering all need to be answered together — Hikube answers all three at once.

---

Looking for more European sovereign cloud alternatives? See also: OVHcloud vs Google Cloud, Hetzner vs AWS, and Scaleway vs Google Cloud.

Frequently Asked Questions

Where is Hidora based and how does that affect jurisdiction?

Hidora SA is headquartered in Geneva, Switzerland, registered under CHE-286.910.173 and listed on the Swiss commercial register (Zefix). Founded in 2016 by CEO Matthieu Robin. All Hikube infrastructure runs in three Swiss data centres — Geneva, Gland, and Lucerne — with native synchronous replication across all three; Hidora states that no customer data transits through a foreign server. Switzerland is not a member of the European Union but has an EU adequacy decision under GDPR, and Swiss federal data protection law (revised nFADP in effect since September 2023) is aligned with GDPR. Crucially, Switzerland is not subject to the US CLOUD Act, and the Swiss constitutional framework provides strong protection against foreign extraterritorial reach. Hidora holds Swiss Hosting certification — the industry mark verifying exclusively Swiss operations.

How does Hikube compare to AWS on GPU workloads?

Hikube publishes a complete GPU-as-a-Service price list: NVIDIA L4 at 165 CHF/month, L40S at 760 CHF/month, A100-80 at 1,495 CHF/month, RTX 6000 Pro at 1,550 CHF/month, H100 at 1,999 CHF/month, and H200 at 2,499 CHF/month (on demand). This spans the top tier NVIDIA GPUs AWS offers via P5 (H100), P4d (A100), G6 (L4). For teams whose workload fits within Swiss capacity, Hikube provides an equivalent hardware tier in Swiss jurisdiction — and with predictable monthly pricing that avoids the AWS complexity of Reserved Instances plus Savings Plans plus Spot. AWS has larger inventory depth and geographic distribution; Hikube has published pricing without an enterprise-sales negotiation layer.

What is KCSP and why does it matter?

KCSP is the Kubernetes Certified Service Provider programme run by the CNCF (Cloud Native Computing Foundation). It is a formal certification that a service provider has demonstrated significant Kubernetes engineering experience across multiple production customer engagements. Only a limited number of providers hold KCSP status — it is a stronger signal than generic 'Kubernetes-managed' marketing claims. Hidora is listed as a KCSP in the CNCF landscape (verifiable at https://landscape.cncf.io/?item=special--kubernetes-certified-service-provider--hidora-kcsp) and is a CNCF Silver member. For teams whose workloads depend on Kubernetes as a first-class platform, this indicates the operator has the actual engineering depth to run it well.

Is Hikube's stack really 100% open source?

Yes. Hidora publishes the Hikube stack composition: Kubernetes on Talos as the Kubernetes distribution, KubeVirt for the VM layer, and SeaweedFS for S3-compatible object storage. The managed database portfolio uses standard open-source engines (PostgreSQL, MySQL, Redis, Kafka, RabbitMQ). The IaC and observability layer is standard OSS (Terraform, Helm, ArgoCD, Prometheus, Grafana). This is meaningfully different from AWS, whose flagship services (DynamoDB, S3, Lambda, EventBridge) are proprietary with no open-source equivalent that runs the same code. For organisations where 'no proprietary lock-in' is a strategic requirement, Hikube's open-source posture is a structural advantage — a Helm chart deployed on Hikube runs unchanged on any other Kubernetes.

What are the certifications relevant for Swiss finance and healthcare?

Hidora lists ISO/IEC 27001 certification (information security management, certified by SQS), the Swiss Hosting label (verifying exclusively Swiss operations) and the Swiss Label. FINMA, the Swiss Financial Market Supervisory Authority, is a regulator and does not certify cloud providers; Hikube instead offers a dedicated Finance & FINMA solution aimed at Swiss banks, insurers and financial-services organisations moving regulated workloads off US hyperscalers. Combined with KCSP status and CNCF Silver membership, this positions Hikube for regulated Swiss workloads, but check the specific attestations your auditors require (for example PCI DSS, which Hikube does not list).

How does pricing compare — with the actual published numbers?

Cloud instances start at 24 CHF/month (s1.small, 1 vCPU / 2 GB); the u1 family starts at 40 CHF/month (1 vCPU / 4 GB, u1.medium) with a linear 1:4 vCPU/RAM ratio scaling to 1,280 CHF/month for 32 vCPU / 128 GB (u1.8xlarge). Managed Kubernetes control plane is 32 CHF/month. Managed databases start at 4 CHF/month (nano tier). S3 object storage is 0.02 CHF/GB/month (0.03 CHF/GB/month for encrypted). GPUs range 165-2,499 CHF/month across L4/L40S/A100/RTX 6000 Pro/H100/H200. Commitment discounts apply on 1/3/5 year terms. Critically, egress traffic — both inbound and outbound — is free at every tier, with no exit fees. Compared with AWS list prices for equivalent compute plus egress-heavy workloads, Hikube is typically 30-60% cheaper on total cost of ownership; on very-high-egress workloads the difference is dramatic because AWS charges per-GB egress and Hikube does not.

Can I run production Kubernetes at scale on Hikube?

This is Hikube's core competency. KCSP-grade Managed Kubernetes with three-DC synchronous replication, S3-compatible storage backing, integrated managed databases, GPU nodes for AI workloads, and standard tooling (Terraform, Helm, ArgoCD, Prometheus, Grafana) support production Kubernetes at scale. The 100% open-source stack means workloads are portable — you can move a Helm chart from Hikube to any other Kubernetes environment without rewriting. Managed Kubernetes control plane is 32 CHF/month.