Conditional verdict — pick the family before you trust “pause”
UpCloud earns a place on AiXVPS when you want a MaxIOPS-forward EU/global hourly prepaid cloud: choose Starter, Premium, or Cloud Native on purpose, debit a prepaid balance by the hour with a 28-day monthly display cap, and treat shut down as a power state—not a savings plan. It is the wrong shortlist if you only want Contabo-style RAM-per-dollar bulk, a RackNerd annual special, InterServer linear monthly simplicity, Vultr’s multi-family/API ladder already decided, ScalaHosting managed WordPress/SPanel responsibility, or Kamatera’s a-la-carte CPU types A/B/D/T.
Who this usually fits
- Operators who want Premium’s bundled MaxIOPS story for always-on n8n, WordPress, or app hosts—and will delete when done.
- Buyers who need EU/global DCs with a prepaid balance model and zero-cost egress marketing under a published Fair Transfer Limit.
- Experimenters who will use Cloud Native to stop compute when powered off—and still delete disks, IPs, and leftover backups.
Who should start elsewhere
- Contabo — always-on RAM-per-dollar self-managed bulk (shape contrast only; no invented winner).
- RackNerd — recurring annual specials.
- InterServer — linear monthly U.S. KVM slices.
- Vultr — composable families, API-heavy networking, evidenced GPU paths already preferred.
- ScalaHosting — managed WordPress/SPanel with intro→renew prepaid clarity.
- Kamatera — CPU-type configurator (A/B/D/T) and monthly/hourly dual mode.
Early warnings. Starter and Premium still bill when shut down. Cloud Native can stop compute when powered off, but storage and IPs continue while allocated. Deleting a server can leave on-demand backups billing. Homepage trial marketing (€200 / 30 days on capture) conflicts with docs (€10 / 7 days plus quotas)—verify the live signup offer. Research froze EUR price tables; recheck USD before quoting dollars. Simple Backup on current plans is paid (+10% to +60% of plan price)—do not promise free Day backups from discontinued General Purpose guides.
AiXVPS publishes UpCloud as a decision cluster, not a scoreboard. The useful outcome is a written family, plan size, datacenter, Simple Backup tier, prepaid balance habit, and delete date—not a vibe that European cloud is “fast.” If you cannot name those fields, you are not ready to deploy.
Think in branches, not slogans. If your job is always-on automation, Premium’s shut-down-still-bills rule should change how you schedule experiments. If your job is a weekend batch, Cloud Native’s stop-compute story only helps when leftover disks, IPs, and on-demand backups are on the same checklist as the VM. If your job is “cheapest possible root,” Contabo’s cash shape may be the honest comparison—and that comparison still does not invent a winner for every workload.
Keep recovery boring. Application-consistent exports beat MaxIOPS marketing when a restore is due. Simple Backup is a paid rotating snapshot layer with leftover billing after delete—not a substitute for n8n JSON, database dumps, or WordPress files. GPU Servers exist as an exit ramp for local inference; they are not the default five-article promise, and Spot carries SLA exclusions.
Decision cards: setup · AI apps · n8n · billing / backup / cancel · all providers.
Prepaid hourly cash-shape (28-day cap, not calendar-month surprise)
UpCloud’s useful money story is prepaid balance debit by the hour, not a classic cPanel monthly invoice alone and not Kamatera’s monthly-prepaid vs hourly-post-billed dual mode.
| Topic | Official shape | Operator habit | Trap |
|---|---|---|---|
| Prepaid balance | Add credits; services deduct hourly | Keep auto top-up or calendar top-ups if always-on | Credits generally non-refundable except documented exceptions |
| 28-day monthly display | Plan estimates use max 28 billable days/month | Compare display rows fairly across calendar lengths | Treat as display estimate—not a private invoice PDF |
| Stop spend | Delete services (pricing FAQ) | Shut down ≠ delete on Starter/Premium | Native: power-off stops compute; disks/IPs remain |
| Trial surfaces | Docs: 7 days + €10 credit + quotas; homepage: €200 / 30 days marketing; FAQ: amount changes | Verify live signup; list docs quotas before planning GPU/Windows | Unused trial credit expires; services removed if no minimum deposit |
Trial conflict—keep both surfaces honest
Docs describe a standard free trial of 7 days from activation with about €10 credit, activation within 365 days of registration, card authorization of 0/1 €/$, quotas such as 2 cores / 4 GB memory total (no Windows), Block Storage totals, Simple Backup available, and GPU / Private Cloud / Windows unavailable. Homepage marketing on the research capture advertised €200 in cloud credits valid for the first 30 days. Pricing FAQ says the free-credit amount changes—see signup. Do not freeze one figure as evergreen. Write the live signup offer into your private ledger.
First deposit money-back is a support path
Docs state a minimum one-time deposit (about 10 in chosen currency) upgrades the account, and the first payment includes a 30-day money-back path up to €500 / $500 via Customer Support. That is not a silent cancel button and not a promise that every top-up is refundable. TOS language generally treats Credits as non-refundable unless UpCloud decides otherwise.
Private ledger fields: family, plan size, DC, storage tier (if Native), Simple Backup tier, estimated €/mo display, Fair Transfer TB row, prepaid balance habit, planned delete date.
MaxIOPS, Fair Transfer, and backups (not footnotes)
MaxIOPS is the Premium production differentiator
Premium marketing centers bundled MaxIOPS storage; Starter uses Standard SSD. Vendor materials describe MaxIOPS as high-performance clustered flash (marketing/docs also cite up to about 100k IOPS read class). Treat performance ceilings as vendor claims—not AiXVPS lab results. Attachable MaxIOPS / Standard / Archive prices appear as EUR/GB/mo rows (Helsinki can differ on Archive)—verify live.
Zero-cost egress under a Fair Transfer Limit
Marketing repeats zero-cost egress. Docs publish Fair Transfer Limit tables by plan code (for example Starter 1×CPU/1GB at 0.5 TB; larger Premium and Native rows scale into many TB). Ingress and private/utility network transfer are included in the documented model. CDN traffic is called out as a separate billing model on the pricing FAQ. Do not invent a soft-cap overage $/GB if the live docs only print limits without a fee formula—verify before promising math.
Simple Backup leftovers after delete
Simple Backup Day / Week / Month / Year adds about +10% / +20% / +40% / +60% of the server plan price on ordinary plan servers. Cloud Native, GPU, and additional disks use per-GB tables. If you disable Simple Backup or delete the server, existing backups convert to on-demand and keep billing until you delete them. Flexible and on-demand backups are separate mechanisms. A “free backups” guide in docs applies to discontinued General Purpose / High CPU / High Memory plans—not current Starter/Premium defaults.
Anti-affinity is HA placement—not a backup
Server groups can enforce strict or preferred anti-affinity across hosts. Useful for dual-node layouts; never a substitute for application-consistent exports.
Scenario → starting envelope → stop or exit signal
Every row is an editorial starting hypothesis. Measure RAM, CPU, disk, and Fair Transfer pressure on your host. Define stop, resize, delete, split, or exit signals before go-live.
| Reader job | Starting hypothesis | Cash / exit note | Stop signal |
|---|---|---|---|
| Docs-style trial host | Starter or small Premium within trial quotas (2 cores / 4 GB total) | Verify live credit offer; export; expect deletion if no deposit | Quota block / port lock / GPU temptation |
| n8n proof (hosted models) | Premium 2–4 GB class | Simple Backup optional; encryption key + JSON export required | Public 5678 / missing key backup |
| Production n8n + Postgres + proxy | Premium 4–8 GB after measurement | Always-on: shut down saves nothing; delete only after exports | OOM / disk fill / leftover on-demand backups |
| Small self-managed WordPress | Premium 2–4 GB or Starter if non-critical | SMTP may need support unlock; not ScalaHosting managed WP | Expecting free malware/plugin coding support |
| Stop-compute weekend batch | Cloud Native + sized disk | Power-off compute; delete disk/IP/backups when finished | Forgotten Native disks |
| Local small-model experiment | Premium high-RAM hypothesis or GPU Servers exit | No tokens/s; Fair Transfer + disk for pulls | Public Ollama bind |
| Dual-node HA | Two Premium/Native + anti-affinity preferred/strict | Backups still separate; LB is another SKU | Anti-affinity as sole recovery plan |
| Always-on RAM-heavy cheap | Compare Contabo first | Contrast RAM/$ vs MaxIOPS/SLA/egress—no invented winner | Forcing UpCloud to be Contabo |
| Need managed WP panel | Prefer ScalaHosting Managed | UpCloud Cloud Servers stay self-managed | Equating Managed DB with SPanel WP Manager |
| Need CPU-type picker | Compare Kamatera | Fixed families ≠ A/B/D/T | Inventing Type T on UpCloud |
| Crypto / spam / unapproved mass mail | Do not recommend | AUP / SMTP defaults | Evasion tips |
UpCloud versus Contabo, RackNerd, InterServer, Vultr, ScalaHosting, Kamatera
Compare cash shape, responsibility, and recovery—not winner rankings.
- Contabo — RAM/$ ladder vs UpCloud MaxIOPS Premium + prepaid hourly + shut-down-still-bills. Shape contrast only.
- RackNerd — annual specials vs UpCloud catalog prepaid families.
- InterServer — linear monthly simplicity vs three-family prepaid + Fair Transfer tables.
- Vultr — multi-family + destroy-to-stop discipline vs UpCloud Native stop-compute + Starter/Premium shut-down trap; map Vultr destroy → UpCloud delete.
- ScalaHosting — managed WP/SPanel + intro→renew cliff vs UpCloud self-managed Cloud Servers.
- Kamatera — a-la-carte CPU types A/B/D/T vs UpCloud fixed Starter/Premium/Native + MaxIOPS brand.
Worksheet fields: cash shape, who owns ops, recovery proof, region/sovereignty need, app path. No score column.
Fit / no-fit, AUP notes, and next decision
Fit
- You will delete leftovers (servers, on-demand backups, Native disks, extra IPs).
- You accept that Starter/Premium shut down still bills.
- You want MaxIOPS production storage or Native modularity on purpose.
- You will verify the live trial offer instead of freezing homepage marketing.
No-fit
- You expect pause = $0 on Premium/Starter.
- You need ScalaHosting-style managed WordPress panel life.
- You need Kamatera-style CPU type threads/cores picker.
- You refuse export-before-delete.
- You want annual-special cash as the primary thesis.
- You treat homepage €200 credits as a guaranteed forever rate card.
AUP and SMTP notes
Illegal and malicious use is prohibited. SMTP port 25 is blocked by default until support unlock after best-practice review—no evasion tips. Standalone AUP HTML was Cloudflare-blocked in research; attribute carefully and recheck live legal pages at draft time.
Next: setup for Control Panel selections; n8n for DIY Compose; AI apps for model placement; billing for export→delete. Also n8n use case, local AI, WordPress, total cost tool.
How to read UpCloud pricing without fooling yourself
Use the pricing page and Control Panel as inputs to a private ledger, not as proof that your invoice will match a screenshot forever.
- Always name family before comparing euro/mo rows.
- Label the 28-day monthly display estimate explicitly.
- If you need USD, re-capture the USD toggle—do not convert EUR with a homemade FX guess for publication.
- Add Simple Backup %, extra disks, extra IPv4, and Native storage lines before you call a shape “cheap.”
- Write Fair Transfer TB for the plan code; leave overage fee UNKNOWN unless live docs print one.
- Trial: paste the live signup sentence into your notes beside the docs 7-day / €10 quota list.
Worked display examples (EUR capture only—recheck live)
These are illustrations from the 2026-11-06 EUR pricing capture, not invoices and not performance promises:
- Starter 2 GB / 1 CPU / 20 GB ≈ €6/mo display — still bills when shut down.
- Premium 4 GB / 2 CPU / 50 GB MaxIOPS ≈ €26/mo display — production hypothesis starting point for small n8n/WP.
- Premium 8 GB / 4 CPU / 100 GB MaxIOPS ≈ €52/mo display — common “growing automation” envelope before measurement.
- Cloud Native 4 GB / 1 CPU ≈ €12/mo compute display + separate storage/IP — stop-compute lab only if you will delete leftovers.
Operational realities buyers underestimate
- Shut down theater. Stopping Premium overnight feels thrifty and still drains prepaid balance.
- Leftover backups. Deleting a server without deleting converted on-demand backups is a classic silent drain.
- Trial optimism. Homepage credit marketing and docs quotas are both official surfaces—planning GPU on a docs trial fails.
- SMTP. Port 25 blocked by default until support unlock after best-practice review.
- EU story ≠ automatic compliance. Datacenter choice and your application controls still matter; read Privacy Notice and shared-responsibility docs for your case.
- DIY n8n. No captured official one-click means you own Compose, keys, and restore proof—or you choose another host with a packaging story you accept.
- Anti-affinity ≠ backup. Spreading VMs across hosts does not replace application-consistent exports.
- Balance alarms. Configure notifications / auto top-up deliberately; the credits-run-out ladder can shut servers down and later delete storages while leftovers still bill.
Sources and updates
Primary public sources for this cluster (dated research capture 2026-11-06 Asia/Shanghai): UpCloud pricing page, Cloud Servers product page, plans/configurations docs, free-trial and account-balance docs, network-transfer Fair Transfer tables, block-storage backups docs, Terms of Service, Privacy Notice, homepage (trial marketing as conflict only). Many /global/ marketing URLs returned Cloudflare challenges from research egress and were not used as evidence.
Update habit. Recheck live pricing (EUR/USD), trial signup offer, Fair Transfer tables, and Simple Backup percentages before you act. Affiliate commission rates are UNKNOWN publicly and omitted from customer-facing body copy.