IONOS Cloud pricing explained

IONOS Cloud pricing is compute cores and RAM per hour billed by the minute, Cubes as flat rates per hour with storage built in, block storage and S3 Object Storage per GB and 30 days, outbound traffic in tiers above 2 TB, and additional static IPv4 addresses; Managed Kubernetes bills only the nodes. On the bill the cores and RAM of the servers that run through the month dominate, followed by block storage; traffic appears only once a contract sends more than 2 TB a month outward.

Pricing dimensions

IONOS does not price servers as fixed types but as the sum of their parts: cores per hour by CPU generation, RAM per GB and hour, storage per GB and 30 days. Everything is usage-based and billed by the minute; all prices are net, with statutory VAT added. The prices here come from the German price page in euro, which shows one list without a location column.

The decomposition into cores, RAM and storage is the trait of this model. A server with two dedicated AMD cores and 4 GB of RAM costs per hour 2 times 0.036 plus 4 times 0.0045, that is EUR 0.09, and EUR 64.80 in 720 hours; the same server with vCPUs 2 times 0.012 plus 4 times 0.002, EUR 0.032 per hour and EUR 23.04 in 720 hours. A Cube Basic S with the same 2 vCPU and 4 GB costs EUR 9.36 per 30 days and brings 120 GB of NVMe storage; Cubes are the flat rate for small, uniform servers, billed by the minute as well, with Windows licences by the minute too. The mechanics of the hyperscalers' instance families are in virtual machines and instance families; IONOS knows no family, only the generation of the core.

Block storage costs per GB and 30 days in three classes that differ by almost a factor of four, and keeps running while the server is stopped; S3 Object Storage costs EUR 0.007 per GB and 30 days without request or retrieval fees, only data transfer outward is billed separately. Outbound public traffic is free per contract up to 2 TB a month and priced above that in four tiers that fall with volume; inbound and private traffic costs nothing, and 1 TB is 1,024 GB. What the allowance means for the bill is in included traffic and flat pricing models. Managed Kubernetes bills the node resources only; the price page lists no line for the control plane, and what a cluster costs beyond that is in containers and Kubernetes.

Invoices, exports and APIs

IONOS issues one invoice per contract and month as a PDF, available in the control panel under Billing and Invoices; prices are net with statutory VAT added, and how it is shown for business customers with a VAT ID in another EU state is in VAT and reverse charge.

The Data Center Designer shows under Management and Cost & Usage the usage and the estimated cost per product category with price per unit, quantity and amount, for the running month or one of the last six months; virtual data centres are aggregated and can be expanded, and products without a mapping sit under Others. The view is refreshed at least daily, so it lags by up to 24 hours, uses list prices without discounts and without VAT, and is, by the documentation's own words, an estimate that may differ from the invoice. Contract owners and contract administrators have access.

The Billing API under api.ionos.com/billing/{contract} delivers the same data machine-readably: invoices per contract, filtered by period; utilization over all periods, per month (YYYY-MM) or per day with an array of data centres by UUID and the meters beneath them; usage with quantities per meter; products as the price list and the profile; plus cost alerts under usage-alert. Authentication is by username and password or by a Bearer token from the DCD's Token Manager, which is mandatory on accounts with multi-factor login. The data is refreshed at least every 24 hours. A line per resource and hour with a price therefore does not exist; the daily utilization per data centre and meter is the finest level, and what it yields for a mapping onto FOCUS columns is in collecting cost data from European providers.

Levers

Configuration

The choice between dedicated cores, vCPUs and Cubes is the largest lever: for the same two cores and 4 GB a factor of almost seven separates the dedicated variant, the vCPU variant and the Cube. Because cores and RAM are priced apart, rightsizing works per dimension here: a server with too much RAM pays for the RAM, not for a larger type. The CPU generation is a price of its own, and the oldest generation is the cheapest. Block storage in the Essential class costs a good quarter of the Performance class; what the class does not need belongs in the cheaper one.

Purchase option

The Cloud Savings Plan lowers the prices of dedicated cores and RAM against a commitment over one or three years; it is booked in the Data Center Designer and requires accepting the purchase conditions. It applies to the resources, not to a server, and so follows the logic of the hyperscalers' savings plans, whose coverage and utilisation commitment coverage and utilisation describes. Cubes and vCPU servers sit outside the plan.

Architecture

The traffic allowance applies per contract, not per server, and the tiers fall with volume: whoever delivers a lot pays less per GB on average, whoever delivers little pays nothing. Private traffic is free, so servers connect over private LANs instead of public IPs; data transfer and egress explains the mechanics. A second static IPv4 costs EUR 5.00 per 30 days, and a load balancer or a NAT gateway saves it.

Where it goes wrong

• The server is stopped, the storage runs. Symptom: a switched-off server keeps causing cost. Cause: block storage and additional IPv4 addresses are billed per 30 days for as long as they exist; only cores and RAM end with the stop.
• The estimate does not match the invoice. Symptom: Cost & Usage in the DCD names a different amount than the invoice. Cause: the view uses list prices without discounts and lags by up to 24 hours; the invoice is binding.
• The RAM outlier. Symptom: a server with few cores costs unexpectedly much. Cause: RAM is billed per GB and hour; 64 GB cost more than two cores.
• A generation by accident. Symptom: two equally sized servers cost differently. Cause: the core price depends on the CPU generation chosen at creation.
• Traffic in the first tier. Symptom: a traffic line at EUR 0.03 per GB. Cause: the contract has passed 2 TB; the allowance is counted per contract, not per server.
• Static IPs without servers. Symptom: IPv4 lines per 30 days with no assignment. Cause: reserved static addresses are billed until released.
• Token expired. Symptom: the Billing API answers with 401. Cause: with multi-factor login a Bearer token from the DCD is mandatory, and it expires.

Example

List prices from the German price page in euro net, read on 2026-10-21: a dedicated AMD EPYC core (Gen 3) costs EUR 0.036 per hour, an Intel Xeon core (Gen 6) EUR 0.046, RAM EUR 0.0045 per GB-hour; a vCPU EUR 0.012 and its RAM EUR 0.002 per GB-hour; a Cube Basic S with 2 vCPU, 4 GB and 120 GB EUR 0.013 per hour or EUR 9.36 per 30 days; block storage Essential EUR 0.04, Balanced EUR 0.07 and Performance EUR 0.15 per GB and 30 days; S3 Object Storage EUR 0.007 per GB and 30 days; outbound traffic above 2 TB EUR 0.03 per GB; a further static IPv4 EUR 5.00 per 30 days. The figures show that two dedicated cores with 4 GB cost EUR 64.80 in 720 hours, the same configuration with vCPUs EUR 23.04 and as a Cube EUR 9.36 including 120 GB of storage, that 100 GB of Performance block storage at EUR 15.00 cost more than a small Cube, and that 3 TB of outbound traffic pay for only 1 TB in the first tier, EUR 30.72 at 1,024 GB per TB.

Sources

Read on 2026-10-21. Prices: IONOS Cloud prices, German page, IONOS Cloud prices, English page in US dollars. Billing: Cost & Usage in the DCD, Billing API, Billing API v3 reference, Cost alerts.