Hetzner Repriced Its Cloud Today, And Some Lines More Than Doubled
On June 15, 2026 at 08:00 CEST, Hetzner put its latest cloud price adjustment into effect. The official figures are striking, and unlike the across-the-board increase in April, this one is deeply uneven. The dedicated vCPU and AMD shared vCPU lines roughly doubled to nearly tripled, while the ARM and Intel-shared lines rose around 30 percent.
This is Hetzner's fourth pricing action of 2026, after setup fees in February, a portfolio-wide 30 to 37 percent increase on April 1, and dedicated setup fees on April 29.
The Numbers That Matter
These are the new monthly prices, in euros excluding IPv4, from Hetzner's official price-adjustment page (Germany and Finland regions), with the increase over the prior price.
Dedicated vCPU (CCX), the hardest hit:
- CCX13: 15.99 to 42.99 (+169 percent)
- CCX23: 31.49 to 85.99 (+173 percent)
- CCX33: 62.49 to 138.49 (+122 percent)
- CCX53: 249.99 to 533.49 (+113 percent)
- CCX63: 374.49 to 853.49 (+128 percent)
AMD shared vCPU (CPX):
- CPX22: 7.99 to 19.49 (+144 percent)
- CPX42: 25.49 to 69.49 (+173 percent)
- CPX52: 36.49 to 100.49 (+175 percent)
ARM (CAX) and Intel shared (CX), far milder:
- CAX11: 4.49 to 5.99 (+33 percent)
- CAX41: 31.49 to 40.99 (+30 percent)
- CX23: 3.99 to 5.49 (+38 percent)
- CX53: 22.49 to 29.49 (+31 percent)
If you have seen a figure of nearly 300 percent going around, that overstates it, but by less than you might expect. The worst line, CPX52, is up 175 percent, which means it now costs 2.75 times what it did the day before. For the CCX and CPX families, more than doubling is the rule, not the exception.
Why It Is So Uneven
Hetzner attributes the change to a "massive increase in procurement costs" in the hardware market. Industry coverage ties that squeeze specifically to surging RAM, NVMe SSD, and GPU demand. Either way, the lines that lean on the most expensive resources, dedicated vCPU cores (CCX) and AMD shared cores (CPX), took by far the largest correction. The ARM-based CAX line, which has always been Hetzner's price-performance star, barely moved by comparison. Alongside the repricing, Hetzner standardized its dedicated server catalog into fixed -1, -2, and -3 types and introduced a cheaper -1-Ltd tier built from lower-cost hardware.
Who Is Actually Affected
This is the part the panic headlines skip, and it matters:
- New orders and rescales pay the new prices. Existing cloud instances keep their current price. Orders placed before June 15 but delivered after still get the old price.
- The trap is the rescale. The moment you resize an existing instance up or down, it reprices at the new rate. So an autoscaling setup or a routine bump-the-box habit can quietly move you onto the new, much higher pricing without a fresh order.
What To Do If You Run On Hetzner
We run production Kubernetes on Hetzner ourselves, so this is the checklist we are actually working through:
- Freeze casual rescales on CCX and CPX instances. Any resize reprices the instance. If a CCX or CPX box does not need to change, leave it alone and keep its old rate.
- Audit which lines you are on. If a workload sits on CCX or CPX out of habit, the ARM CAX line now has an even wider price advantage (it only rose about 30 percent) and is an excellent fit for most stateless services and many database workloads.
- Re-run your cost model before you scale. A cluster that was cheap to grow last week may not be. Size new nodes against the new prices, not the old ones.
- Look at the -Ltd tier for dedicated servers. The new limited tier is built specifically to be cheaper and may cover workloads that do not need guaranteed top-bin hardware.
- Re-evaluate single-provider risk. Four price moves in one year is a trend, not a one-off. For the worst-hit workloads it is worth pricing a second provider, even if you decide to stay, just so you know your options.
Right-sizing a fleet against a moving price list, and deciding when to migrate workloads versus absorb the increase, is exactly what our cloud cost optimization and Kubernetes management work covers.
The Bigger Picture
Even after this increase, Hetzner remains cheaper than the major hyperscalers for equivalent raw compute. The story is not that Hetzner became expensive. It is that the gap narrowed sharply on specific lines, and that the era of treating Hetzner pricing as a fixed constant is over. Hardware procurement costs are squeezing every provider, and the ones who started cheapest had the most room to rise. Plan capacity with that in mind, and retire the assumption that scaling up is always cheap.
Sources
Prices and the effective date are from Hetzner's official price-adjustment page and the standardization pressroom announcement (effective June 15, 2026, 08:00 CEST, applying to new orders and rescales only; existing instances keep their price). The April 2026 context (30 to 37 percent cloud increases) is from Hetzner's April statement as reported by Tom's Hardware. The percentage figures above are calculated from Hetzner's published old and new monthly euro prices for the Germany and Finland regions.
Talk to the engineer who will own your stack.
No account managers, no offshore handoff. Senior DevOps, direct. Tell us what you are dealing with and you get a straight answer.
Related News
Azure Returns 410 Gone for GPT-4o on October 1 and Auto-Upgrade Skips the Deployments That Matter
On October 1, 2026, Azure OpenAI in Microsoft Foundry retires the GA gpt-4o (2024-11-20) and gpt-4o-mini (2024-07-18) versions, after which calls to a retired deployment return HTTP 410 Gone. Standard-family deployments are auto-upgraded region by region, but Provisioned (PTU) deployments and anything set to NoAutoUpgrade are not, and Microsoft says the date is not extendable.
CloudStay on EKS 1.33 and AWS Starts Billing You 6x From August
Amazon EKS Kubernetes 1.33 leaves standard support on July 29, 2026, and any cluster still on it is enrolled in extended support by default at six times the standard control-plane rate. The hourly charge rises from 0.10 to 0.60 US dollars per cluster, and the standard patch cadence narrows. Fleets with many clusters feel it first.
CloudAWS Shuts Down App Mesh for Good on September 30 and Your Service Mesh Goes With It
AWS ends all support for App Mesh on September 30, 2026, after which the console and every App Mesh resource go dark and the Envoy data plane is no longer managed. New onboarding has been blocked since September 24, 2024. For ECS, AWS points to Service Connect; EKS teams have their own path.