Linode (Akamai Cloud) vs. Hetzner Cloud
Linode and Hetzner appeal to a similar kind of buyer: developers who want a Linux-first cloud platform without the complexity or cost structure of AWS. But they solve that problem in very different ways.
Linode, now part of Akamai Cloud, has steadily expanded beyond basic VPS hosting into managed databases, Kubernetes, object storage, and higher-end infrastructure such as GPU instances. Hetzner has stayed much closer to the infrastructure layer: low-cost virtual machines, strong European network performance, and unusually generous transfer limits, with far less emphasis on managed services.
That difference becomes especially clear at the 2 vCPU / 4 GB tier. A Linode 4 GB Shared CPU instance costs $24/month, while a Hetzner CPX22 costs €7.99/month, or about $9.49/month after Hetzner’s April 2026 price adjustment. On paper, the specs look similar. In practice, the price gap is large enough that the real question is not whether Hetzner is cheaper, but what you give up by paying less.
To answer that, I provisioned a Linode 4 GB Shared CPU instance in Chicago and a Hetzner CPX22 in Helsinki, ran YABS on both, and compared the results across CPU, disk, network, provisioning, dashboard experience, pricing, and managed services.
Hetzner pricing note: All Hetzner prices in this article reflect the April 1, 2026 adjustment. The CPX22 moved from €5.99/month to €7.99/month in Germany and Finland regions. Hetzner’s US regions also include only 1 TB of transfer, compared with 20 TB in EU regions.
The instances compared:
- Linode: Linode 4 GB Shared CPU (2 vCPU / 4 GB RAM / 80 GB SSD) in Chicago (us-ord) for $24/month
- Hetzner: CPX22 (2 vCPU / 4 GB RAM / 80 GB NVMe SSD) in Helsinki (hel1) for €7.99/month (about $9.49/month)
⚖️ Testing conditions
Linode was benchmarked in Chicago (us-ord) on a fresh Ubuntu 24.04 LTS install. Hetzner was benchmarked in Helsinki (hel1) on the same OS. Both ran YABS v2025-04-20 with no other software present. The different origins mean transatlantic latency numbers reflect geography as much as network peering — Chicago sits roughly 100 ms from Amsterdam, Helsinki roughly 30 ms. CPU and disk figures don't share that caveat; they reflect the hardware regardless of where the server is located.
| Linode (Akamai Cloud) | Hetzner Cloud | |
|---|---|---|
| Plan tested | Linode 4 GB Shared CPU, $24/month | CPX22, €7.99/month (~$9.49) |
| vCPU | 2 | 2 |
| RAM | 4 GB | 4 GB |
| Storage | 80 GB SSD | 80 GB NVMe SSD |
| Transfer included | 4 TB | 20 TB (EU) |
| Region tested | Chicago (us-ord) | Helsinki (hel1) |
| CPU hardware | AMD EPYC 7713 | AMD EPYC-Rome @ 2.4 GHz |
| Geekbench 6 single core | 1,343 | ~939 |
| Geekbench 6 multi core | 2,490 | ~1,672 |
| Disk 4k IOPS (combined) | 94.5k | ~40.9k |
| Disk sequential (1m) | 5.54 GB/s | ~2.97 GB/s |
| IPv6 by default | Yes | Yes |
| Disk encryption by default | Yes | No |
| SLA (compute) | 99.99% | Not specified |
| Managed databases | Yes ($81.60/month+) | No |
1. Performance benchmarks: Point Linode
CPU
Linode's Shared 4 GB runs AMD EPYC 7713 — a Milan-generation server processor, released in 2021. Hetzner's CPX22 runs AMD EPYC-Rome at 2.4 GHz, the prior generation from 2019. That two-year hardware gap is visible in benchmarks.
Linode (Chicago, AMD EPYC 7713):
| Test | Score |
|---|---|
| Single core | 1,343 |
| Multi core | 2,490 |
| Full result | https://browser.geekbench.com/v6/cpu/16880203 |
Hetzner (Helsinki, AMD EPYC-Rome @ 2.4 GHz):
| Test | Score |
|---|---|
| Single core | ~939 (avg of two runs: 964 / 914) |
| Multi core | ~1,672 |
| Run 1 | https://browser.geekbench.com/v6/cpu/16879506 |
| Run 2 | https://browser.geekbench.com/v6/cpu/16879916 |
Linode's 1,343 single-core result is 43% faster than Hetzner's ~939 average. The two Hetzner runs show 50-point variance (964 vs. 914), which is expected on shared infrastructure — shared VMs don't always land on the same physical core state, and the CPX series is explicitly a shared product. Taking the lower run, the gap widens to 47%.
For a single-threaded workload — a web server handling requests, a compilation job, a Redis instance — that 43–47% spread is real. For workloads that are memory-bound, network-bound, or spend most of their time waiting, it matters considerably less.
Disk I/O
The storage gap is wider than the CPU gap:
| Block size | Linode (Chicago) combined | Hetzner (Helsinki) combined |
|---|---|---|
| 4k | 94.5k IOPS | ~40.9k IOPS |
| 64k | 56.0k IOPS | ~27k IOPS |
| 512k | 7.8k IOPS | ~3.5k IOPS |
| 1m | 5.54 GB/s | ~2.97 GB/s |
Linode's 4k combined result of 94.5k IOPS is 131% higher than Hetzner's ~40.9k. Sequential throughput at 1m block size is 87% faster: 5.54 GB/s versus ~2.97 GB/s. The disk story here matters more than the headline numbers. Hetzner's CPX22 runs NVMe storage in a local RAID10 on the host — fast hardware by shared-tier standards. Linode's Shared CPU plan uses regular SSD storage, not NVMe, yet benchmarks 131% faster on 4k IOPS and 87% faster sequentially. The gap likely reflects Linode's storage backend being less contended at the time of testing rather than a hardware advantage, though sustained workloads under load may show different results. For database workloads, log-intensive applications, or anything where disk latency compounds across requests, this difference is worth noting.
Network
The Chicago vs. Helsinki origin makes it hard to compare transatlantic results directly, but each platform's throughput to its nearest test nodes tells you what the upstream peering looks like.
Linode (Chicago, us-ord):
| Provider | Location | Send | Receive | Ping |
|---|---|---|---|---|
| Clouvider | London, UK (10G) | busy | 2.00 Gbits/sec | 89.1 ms |
| Eranium | Amsterdam, NL (100G) | 2.26 Gbits/sec | 2.38 Gbits/sec | 99.7 ms |
| Leaseweb | Singapore, SG (10G) | 732 Mbits/sec | 641 Mbits/sec | 221 ms |
| Clouvider | Los Angeles, CA (10G) | 2.42 Gbits/sec | 4.14 Gbits/sec | 52.2 ms |
| Leaseweb | NYC, NY (10G) | 4.34 Gbits/sec | 7.98 Gbits/sec | 20.1 ms |
| Edgoo | Sao Paulo, BR (1G) | 1.64 Gbits/sec | 1.59 Gbits/sec | 162 ms |
Hetzner (Helsinki, hel1):
| Provider | Location | Send | Receive | Ping |
|---|---|---|---|---|
| Clouvider | London, UK (10G) | 5.02 Gbits/sec | 5.22 Gbits/sec | 38 ms |
| Eranium | Amsterdam, NL (100G) | 8.08 Gbits/sec | — | 28.8 ms |
| Leaseweb | Singapore, SG (10G) | — | — | ~190 ms |
| Clouvider | Los Angeles, CA (10G) | — | — | ~157 ms |
| Leaseweb | NYC, NY (10G) | 1.73 Gbits/sec | — | 131 ms |
This is where the geography matters most. Hetzner's Helsinki node pushes 8.08 Gbits/sec to Amsterdam at 28.8 ms and 5+ Gbits/sec to London — exceptional European peering. Its North American results are predictably worse: 1.73 Gbits/sec to NYC with 131 ms latency. Linode from Chicago shows the inverse: 7.98 Gbits/sec receive from NYC at 20.1 ms, with transatlantic performance settling at 2.26–2.38 Gbits/sec to Amsterdam at ~100 ms.
If your users are in Europe, Hetzner's network profile fits better than Linode's Chicago node. If they're in the US, Linode wins. Both platforms have multiple regions, so you can deploy closer to your audience — but within this benchmark's datacenter selection, the network stories are mirror images of each other.
2. Provisioning experience: Point Hetzner
Hetzner's Cloud Console provisioning page runs top to bottom on a single screen with no branching tabs or multi-step flow: choose the plan family and tier, pick a location, select an OS image, configure networking, add an SSH key, and create. The design makes choices in sequence rather than expanding optionally, which means you're always progressing rather than exploring.
IPv6 is enabled by default alongside IPv4 with no extra step required. Servers are typically ready in under 30 seconds — noticeably faster than either Linode or DigitalOcean. Disk encryption is not enabled by default and requires manual configuration after provisioning.
Linode's provisioning page is organized around tabs at the top — OS, Marketplace, StackScripts, Images, Backups, Clone Linode — which surfaces more creation paths without overwhelming the main form. The plan selector shows all resource details on each card.
Disk encryption is on by default in Linode's Security section, which is a meaningful baseline for teams with compliance requirements. The recommended network interface type ("Linode Interfaces") is currently in beta, which adds a small uncertainty for production use. Provisioning takes 60–90 seconds — longer than Hetzner's sub-30 second window.
Both flows are straightforward. Hetzner's edge is speed and simplicity at the cost of surfacing fewer options. Linode's tab structure is more flexible for teams with existing automation or template workflows (StackScripts, cloning), and disk encryption by default is a practical advantage.
3. Dashboard and management: Point Linode
Hetzner's Cloud Console server detail view shows status indicators, IP addresses, usage graphs (CPU, disk I/O, network, traffic usage), and a right-side action panel for power operations, snapshots, rescue mode, and rebuild. It's efficient and uncluttered, with no unnecessary chrome.
What the Hetzner console lacks is account-level organizational tooling. There's no project or team permission model, no resource tagging with role-based access controls, and no equivalent of multi-user fleet management. For a single developer managing a handful of servers, this is irrelevant. For a team running multiple environments across a shared account, the absence of permission scoping becomes a real operational gap.
Linode's Cloud Manager addresses this directly. The detail view covers power controls, performance graphs, Lish browser-based serial console access, snapshot management, and resize — and at the account level, VPC, Cloud Firewall, Placement Groups, DNS Manager, and team-scoped access controls are all present and included at no extra cost.
The Lish console — Linode's browser-based serial access — is worth calling out specifically. It works when SSH is unreachable (misconfigured firewall, locked-out root, failed network configuration), supports SSH-to-Lish as an out-of-band path, and has been reliable for years. Hetzner offers a console access option via the Cloud Console, but it's less capable as an out-of-band recovery tool.
The Akamai rebrand adds some navigational friction — some screens say "Linode," others say "Akamai Cloud" — but this is a cosmetic issue rather than a functional one.
4. Ecosystem and managed services: Point Linode
Hetzner's product catalog outside of compute is short by design: managed Kubernetes (HKE, free control plane), load balancers, block volumes, object storage (EU-only), and a basic firewall. That's the full list. There are no managed databases. No PaaS or App Platform equivalent. No container registry. No serverless. No DNS with integrated record management for external zones.
Linode's ecosystem covers the same compute foundations and extends significantly further:
| Service | Linode | Hetzner |
|---|---|---|
| Managed PostgreSQL | $81.60/month (dedicated G7) | ✗ |
| Managed MySQL | $81.60/month (dedicated G7) | ✗ |
| App Platform (PaaS) | ✓ | ✗ |
| Kubernetes control plane | Free (standard) / $60/month (HA) | Free |
| Object storage | $5/month, 250 GiB, no CDN | €6.49/month base, EU only, no CDN |
| Block storage | $0.10/GB/month | €0.057/GB/month |
| Load balancer | $10/month (NodeBalancer) | €8.91/month |
| DNS Manager | ✓ (free) | ✗ (basic, no external zone management) |
| Serverless | ✓ (Akamai Functions) | ✗ |
| GPU instances | ✓ (including Blackwell) | ✗ |
Databases
The absence of managed databases on Hetzner isn't a minor gap — it means any database that needs to run on the platform requires self-managed installation, patching, backup configuration, and failover handling. For teams already comfortable with Postgres administration, this is fine. For teams that want to spin up a database alongside their application without taking on database operations, Linode is the only option. The $81.60/month entry price for Linode's managed PostgreSQL (single node, dedicated G7 hardware) is steep compared to other providers, but it at least exists.
Object storage
Both platforms offer S3-compatible object storage starting near $5/month, but Hetzner's service is EU-only — not available from US Hetzner regions. Neither includes a bundled CDN.
GPU and AI
Akamai's post-acquisition investment in GPU infrastructure shows up here. Current Linode GPU instances include the NVIDIA RTX PRO 6000 Blackwell Server Edition — 96 GB VRAM per card, single-card instances from $1,665/month — with Akamai citing up to 1.63x the inference throughput of an H100 under FP4/FP8 workloads. For teams building inference pipelines or running large model workloads, this is meaningfully different from a generic cloud GPU offering. Hetzner offers nothing in this category.
5. Pricing: Point Hetzner
The raw price comparison is straightforward: Hetzner CPX22 at €7.99/month (~$9.49) versus Linode 4 GB at $24/month — a 60% cost reduction for identical on-paper specs (2 vCPU / 4 GB / 80 GB). That gap holds at every tier on the CPX series:
| Hetzner plan | Spec | Price (post-April 2026) | Equivalent Linode plan | Linode price |
|---|---|---|---|---|
| CPX22 | 2 vCPU / 4 GB / 80 GB | €7.99/month | Linode 4 GB | $24/month |
| CPX32 | 4 vCPU / 8 GB / 160 GB | €13.99/month | Linode 8 GB | $48/month |
| CPX42 | 8 vCPU / 16 GB / 320 GB | €25.49/month | Linode 16 GB | $96/month |
The transfer gap amplifies the value story. Hetzner European plans include 20 TB/month outbound at no extra cost; Linode includes 4 TB. At 20 TB, Linode's overage would add $80/month at $0.005/GB — turning a $24/month plan into $104/month for a bandwidth-heavy workload. Hetzner's EU overage rate is €1.00/TB. For bandwidth-intensive deployments — media hosting, software distribution, data pipelines with heavy egress — Hetzner's transfer economics are substantially better.
A few Hetzner caveats worth noting. US region plans include only 1 TB transfer rather than 20 TB — Hetzner's transfer advantage is specific to EU deployments. Block volumes (€0.057/GB/month) and load balancers (€8.91/month) are slightly cheaper than Linode's equivalents ($0.10/GB/month and $10/month respectively). Object storage (€6.49/month base) is marginally more expensive than Linode's $5/month minimum but EU-only. Snapshots cost €0.014/GB/month on Hetzner; Linode charges $0.10/GB/month for stored images.
Add-ons Linode charges for that Hetzner includes at no cost: cloud firewall (included on Hetzner). Add-ons Linode includes that Hetzner charges for: nothing equivalent on this comparison — Hetzner simply doesn't offer the services (managed databases, GPU, PaaS) to charge for.
6. Documentation: Point Linode
| Linode | Hetzner | |
|---|---|---|
| Infrastructure depth | ✓✓ | ✓ |
| CLI / API coverage | ✓✓ | ✓✓ |
| Beginner onboarding guides | ✓ | ✓ |
| Community tutorials | ✓ | partial (subreddit/LowEndTalk) |
| Real command output in guides | ✓✓ | ✓ |
Hetzner's official documentation at docs.hetzner.com covers what the platform does — server provisioning, networking, Kubernetes, API usage, the CLI (hcloud) — with accuracy and appropriate technical depth. It's documentation designed for people who already know what they're doing. There's no tutorial library of the kind Linode maintains, and the community knowledge base is distributed across the Hetzner subreddit, LowEndTalk, and third-party guides rather than consolidated under a first-party URL.
Linode's documentation at techdocs.akamai.com has accumulated over two decades of Linux infrastructure guides. Guides include expected terminal output, cover edge cases in configuration, and go deep on topics like networking, kernel management, and production hardening. For someone running an application on Linode who hits an unfamiliar error, the odds of finding a relevant guide with real commands and expected output are high. Post-acquisition branding drift means some pages reference Linode and others Akamai Cloud — a navigational nuisance but not a gap in coverage.
Final thoughts
This comparison comes down to a simple trade-off: better hardware and managed services on one side, much lower cost on the other.
On raw performance, Linode is clearly ahead. Its CPU scores are meaningfully higher, disk performance is much stronger, and its Chicago test node delivered much better results for US traffic than Hetzner’s Helsinki instance. It also includes features Hetzner does not offer at all, such as managed databases, App Platform, serverless options, and GPU infrastructure. For teams that want more than a VM, Linode is the more complete platform.
Hetzner’s advantage is just as clear: price. Even after the April 2026 increase, the CPX22 costs less than half of Linode’s comparable plan, and for European deployments it includes 20 TB of transfer, which changes the economics completely for bandwidth-heavy workloads. Hetzner is not trying to match Linode’s ecosystem. It is offering a very inexpensive server with good-enough performance and excellent European network positioning.
That makes the choice fairly straightforward. If you need managed services, stronger disk performance, better US network placement, or room to grow into a broader platform, Linode justifies its higher price more easily. If you are comfortable self-managing your stack and your priority is keeping infrastructure costs as low as possible, especially for European workloads, Hetzner is the better value.