Netcup Review 2026: Dedicated Cores Cheaper Than Hetzner, So Why Doesn't Everyone Switch?
Netcup is the German host that gets recommended with a "but." The pitch is strong: dedicated CPU cores, on recent AMD EPYC silicon, for less than Hetzner charges for shared vCPU. The benchmarks back it up. Yet plenty of people who hear that still don't switch, and the reasons are real: a support reputation that bites when something goes wrong, a year-long minimum term, and a 2024 episode that's worth understanding before you trust the "dedicated" label. This review covers both halves honestly.
Performance: the real benchmarks
This is where Netcup earns its reputation, and the numbers are specific, not vibes. The current RS 1000 G12 (4 dedicated cores, 8GB DDR5 ECC, 256GB NVMe) runs on AMD's EPYC 9645 (Zen 5c, "Turin Dense"). Public benchmark runs measured:
| RS 1000 G12 (EPYC 9645) | Measured |
|---|---|
| Geekbench 6 single-core | ~2,200 |
| Geekbench 6 multi-core | ~7,150 |
| Disk 4K random IOPS | ~50k-93k (read) |
| Disk sequential (1M) | ~1.3-1.9 GB/s |
| Memory bandwidth | ~70 GB/s read |
| Network | 2.5 Gbit port (~2.4-2.8 Gbps measured) |
Two independent benchmark runs landed within ~3% of each other, so these are solid. For context, the prior G11 generation scored Geekbench 6 around 1,754 single / 3,876 multi, so G12 is a large generational jump, roughly 30-85% faster multi-core. The disk is healthy NVMe (no Contabo-style throttle on the current line), and crucially the cores are dedicated, which is the whole reason to consider Netcup over a shared-vCPU host. (Figures from public YABS/Geekbench runs on VPSBenchmarks plus a corroborating community review; verify your own with a quick YABS pass.)
"Dedicated cores", and the G11 lesson
The selling point is dedicated cores: unlike Hetzner's shared CPX vCPU (explicitly shared, throttled under sustained load) or Contabo's oversold shared CPU, a Netcup RS core is supposed to be yours, no noisy neighbor stealing cycles. Mostly that holds, and it's the real differentiator. But there's a cautionary tale that keeps the community honest about the word "dedicated."
The G11 oversell episode (2024-2025): on the previous-generation G11 RS line, Netcup oversold, too many "dedicated" vCPUs packed onto nodes. Users reported their boxes losing around half their performance within about a month of ordering (CPU steal plus disk-IO degradation), and the complaints piled up on LowEndTalk until Netcup paused G11 RS orders and moved some customers to other nodes. The lesson isn't "avoid Netcup", it's that "dedicated" is a promise that can strain under commercial pressure, and that the current G12 line is what superseded the problem. The benchmarks above are G12; don't let the G11 saga tar the current generation, but do know the company has overcommitted a "dedicated" line before.
Pricing & the 2026 hikes
The RS 1000 G12 runs about €8.74/mo on an annual EU contract, or around €12.79/mo incl. VAT monthly, on a 12-month minimum term with no setup fee. Per dedicated core, that undercuts Hetzner's shared vCPU pricing, which is the headline value.
But 2026 brought increases you should factor in. Per Netcup's own price-adjustment notice, driven by the year's memory-supply crisis: existing contracts +18.51% (effective next billing period from May 1, 2026), new orders +24.33% (from March 19, 2026), and storage add-ons +21.52%. Netcup offered existing customers a special termination right alongside the change. So the "cheaper than Hetzner" math still holds (Hetzner raised prices the same season), but quote current live figures, the whole budget tier repriced this spring.
Support & the friction
This is the "but" that stops people. Netcup's support draws recurring complaints, the most serious being abuse-related account shutdowns: reports of servers suspended or, in some accounts, deleted over abuse notices with slow or unhelpful responses, sometimes without a clear final warning. (These are community/Trustpilot reports rather than a single documented incident, so weight them as a reputation pattern, not a guarantee of your experience.) On top of that: a 12-month minimum term (no hourly billing like Hetzner's cloud), and a control panel (SCP) the community broadly considers a step behind Hetzner's polished console. None of this touches raw performance, it's the experience around the hardware. Netcup is better silicon, not better hand-holding.
Who Netcup is for
Right for: anyone who wants dedicated-core performance per euro and is comfortable self-managing, a developer running a CPU- or database-heavy workload that needs consistent (not contended) performance, someone migrating off Hetzner's throttled shared vCPU or Contabo's oversold CPU. For these, the G12 RS line is strong value. See current Netcup RS plans →
Wrong for: anyone who wants hourly billing or to spin things up and down freely (the 12-month term kills that), anyone who values a polished console and responsive support over raw specs (Hetzner is the better fit), and anyone running content with any abuse-adjacent risk profile, given the shutdown reports.
FAQ
Is Netcup good?
For performance per euro, yes, dedicated cores on EPYC 9645 (G12), Geekbench 6 ~2,200 single / ~7,150 multi, healthy NVMe disk. The hesitation is support reputation (abuse shutdowns), a 12-month term, and a dated panel. Great hardware, more friction around it.
Netcup vs Hetzner, which is better?
Netcup RS = dedicated cores (consistent performance); Hetzner CPX = shared vCPU that throttles under load. For guaranteed CPU per euro, Netcup. For hourly billing, a better console, and smoother support, Hetzner. See our full comparison.
Did Netcup raise prices in 2026?
Yes, per its official notice: existing contracts +18.51% (from May 1, 2026), new orders +24.33% (from March 19), storage +21.52%, citing the 2026 memory crisis. A special termination right was offered.
What was the Netcup G11 problem?
Netcup oversold the previous-gen G11 RS line in 2024-2025; users saw ~50% performance loss within a month, prompting Netcup to pause G11 orders. The current G12 line superseded it and benchmarks strongly, don't extrapolate G11's issues onto G12.