ChatGPT
Perplexity
Claude
Hetzner raised cloud server prices 30-43% in April 2026, followed by two more pricing events - the latest taking effect June 15. Three price increases in five months. If you’re placing new orders or rescaling existing servers, you’re on the new rates, and the gap between Hetzner and AWS or Azure is narrower than it was last year.
This post breaks down what actually changed, why the increases are structural rather than temporary, and which workloads now make more sense on a managed cloud platform.
How Much Did Hetzner Prices Actually Go Up?
The April 1, 2026 round hit hardest. Specific numbers:
- CX23 (shared vCPU): EUR 2.99 -> EUR 3.99/month (+33%)
- CAX11 (Arm cloud): EUR 3.29 -> EUR 4.49/month (+37%)
- CPX11 (USD billing): $4.99 -> $6.99/month (+40%)
- CCX13 (dedicated vCPU, USD): $14.49 -> $19.99/month (+38%)
- Block volumes (USD): up 53%
- Memory add-ons: up to 575%
A second round in late April adjusted dedicated server setup fees. The third round, effective June 15, standardized the dedicated server lineup into numbered tiers (“-1”, “-2”, “-3”) and introduced a new “Limited” tier for hardware sourced at lower procurement cost. Monthly prices on new dedicated orders went up; setup fees came down. Existing server contracts are protected - but rescaling or adding capacity puts you on new pricing.
OVHcloud, Scaleway, and IONOS raised prices in the same window. The DRAM shortage is an industry-wide problem, not a Hetzner-specific one. Hetzner published a full statement on the June 2026 standardization and pricing changes that covers which products are affected and the new tier naming structure.
Why This Is Happening (and Why It Won’t Reverse Quickly)
DRAM costs increased up to 500% between September 2025 and early 2026. The underlying cause is that AI hyperscalers building GPU clusters pulled manufacturing capacity toward High-Bandwidth Memory - the stacked memory inside H100 and B200-class GPUs. Each gigabyte of HBM consumes roughly three times the wafer capacity of standard DDR5 server DRAM. Memory manufacturers including Samsung, SK Hynix, and Micron reallocated production accordingly because HBM commands a price premium over four times that of conventional server memory.
The supply chain recovery timeline is projected through 2027-2028. More increases are likely before prices stabilize.
Is Hetzner Still Cheap Compared to AWS?
Yes - Hetzner is still significantly cheaper for raw compute. Even at the new rates, you’re paying roughly 2-3x less than DigitalOcean or Linode for equivalent specs, and 5-8x less than AWS EC2 on-demand pricing for similar vCPU and RAM ratios.
If raw compute cost per hour is your only variable, Hetzner still wins.
The question worth asking now is whether raw compute cost per hour is the right variable for your workloads.
When Does Migrating to AWS, Azure, or GCP Make Sense?
Migrating to a hyperscaler rarely reduces your compute bill directly. It tends to save money (and engineering time) on everything around compute.
Your team manages infrastructure that AWS or Azure would manage instead. RDS instead of self-hosted PostgreSQL. EKS instead of Kubernetes installed on bare metal. ElastiCache instead of your Redis setup. Each managed service costs more per unit than the Hetzner equivalent - but removes operational burden that either consumes engineer hours or quietly accumulates risk. At current Hetzner pricing, the cost-versus-ops-burden calculation looks different than it did in 2024.
You need compliance certifications your team doesn’t hold. SOC 2, ISO 27001, HIPAA BAAs, PCI DSS - AWS and Azure hold these at the platform level. Getting equivalent coverage on Hetzner infrastructure means third-party audits, continuous evidence collection, and internal security program investment. For regulated workloads, a hyperscaler is often the cheaper compliance path in total cost.
Your users are outside EU and US regions. Hetzner operates in Germany, Finland, and the US. AWS runs 34 regions globally; Azure operates 60+. If your customer base is in APAC, the Middle East, or LATAM, latency from the nearest Hetzner region is a meaningful product issue. Geographic coverage alone can justify a hyperscaler switch for customer-facing applications.
You have SLA obligations to your own customers. Hetzner doesn’t publish the same availability commitments as AWS or Azure. For internal tools and development environments, that’s fine. For production systems where downtime costs you revenue or breaches your customer contracts, the risk math is different.
What You Actually Gain by Moving
Beyond the compliance and availability points:
Elastic scaling without migration work. On Hetzner, scaling up typically means provisioning a new server, migrating data, and managing a downtime window. On AWS or Azure, you resize an instance or configure auto-scaling policies. For workloads with variable demand - anything with traffic spikes, batch processing, or seasonal patterns - this is a real operational difference that compounds over time.
Managed Kubernetes at the control plane level. EKS, AKS, and GKE handle Kubernetes control plane management, version upgrades, and node group scaling. Running Kubernetes on Hetzner means your team owns all of that. When you factor in the engineering hours to maintain a self-managed K8s cluster, managed Kubernetes on a hyperscaler often closes the per-node cost gap considerably.
Reserved pricing brings AWS/Azure costs down. AWS Reserved Instances and Azure Reserved VM Instances reduce costs 40-60% from on-demand pricing for predictable workloads. Combine with spot or preemptible instances for batch workloads and the raw compute gap with Hetzner narrows to the point where managed services and operational savings often tip the calculation.
Native ecosystem integrations. IAM, CloudWatch, EventBridge, Azure Entra ID, Service Bus - these integrations work natively when your infrastructure lives on the platform. Building equivalent integrations for Hetzner-hosted workloads means maintaining glue code indefinitely. The cumulative maintenance cost is rarely captured in a simple per-hour compute comparison.
How to Approach a Hetzner Migration
If you’re evaluating a move, a structured approach saves significant pain later. The 6 Rs cloud migration framework (Rehost, Replatform, Repurchase, Refactor, Retire, Retain) is the standard way to categorize workloads before touching infrastructure. Not everything should move, and not everything that moves should be rewritten.
A reasonable starting sequence:
- Inventory your Hetzner resources - servers, volumes, snapshots, load balancers, object storage. Most teams find resources they forgot they were paying for.
- Classify each workload by migration path - lift-and-shift for speed on non-critical systems, replatform where managed services reduce operational burden, retire what’s idle or redundant.
- Plan for the common cloud migration challenges that consistently slow teams down: network configuration, DNS cutover timing, data transfer costs (egress from Hetzner is not free), and IAM/access control gaps.
- Run dual-environment during transition - keeping Hetzner live costs money during migration, but a hard cutover without testing is riskier than the overlap cost.
- Validate the numbers post-migration - managed services, reduced ops time, and compliance savings should appear in the cloud migration ROI model within 6-12 months.
For tooling, AWS Application Migration Service (MGN) handles VM-to-EC2 replication with continuous block-level sync and low-downtime cutover. Azure Migrate covers similar ground for Azure targets. A broader cloud migration tools comparison helps if you’re evaluating options before committing to a destination platform.
The Honest Assessment
Hetzner’s pricing increase is a trigger for re-evaluation, not necessarily a verdict that you should leave.
For pure compute workloads with no compliance requirements, no global reach needs, and a strong internal ops team, Hetzner at the new prices is still a defensible choice. You’re still getting good hardware at below-market rates - the gap just closed.
But if you’ve been treating the “Hetzner is so much cheaper” assumption as a settled fact while avoiding a cloud migration evaluation - that assumption is now weaker. When you account for managed services, compliance overhead, and engineer time, many workloads have already crossed the threshold where a hyperscaler is the lower total cost option.
Three price increases in five months, with DRAM supply constraints projected through 2028, means the old Hetzner pricing floor isn’t coming back soon. Building infrastructure cost models around it is a bet worth revisiting.
Planning a Move Away from Hetzner?
The migration planning stage is where most teams either capture significant savings or create expensive problems. Getting the workload classification right before touching infrastructure makes the difference.
Tasrie IT Services provides end-to-end cloud migration services to help you:
- Assess your current Hetzner workloads and map each to the right migration path
- Avoid common migration pitfalls - networking, egress costs, access control, and DNS cutover issues
- Select the right managed services on AWS, Azure, or GCP to replace self-managed infrastructure
- Build a clear cost model before and after migration so the ROI is measurable
We work with teams across the UK and UAE migrating from bare-metal and VPS providers to managed cloud platforms - with minimal downtime and clear cost outcomes.