Google Compute Engine vs Scaleway

Google Compute Engine is a global hyperscaler that bills its VMs by the second in 43 regions. Scaleway is a European cloud in four EU cities that bills its Instances by the hour, which costs less for machines that stay on and more for short jobs.

The companies and products

Google Compute Engine

Google Compute Engine is Google Cloud's general-purpose VM service, not a sandbox product: you rent virtual machines by the second and set them up yourself.

You pick a machine type (E2, N2D, C3D, Arm T2A, GPU families and more), a boot image and a network. Compute Engine boots a KVM-based VM with its own kernel and root access. Disks are Persistent Disk or Hyperdisk volumes, and VPC firewall rules control traffic. You set up the agent parts yourself: agents, a desktop, HTTPS URLs and key handling. Compute Engine serves every kind of server workload, from websites to GPU training, in Google Cloud regions worldwide.

Scaleway

Scaleway is a European cloud and AI provider, part of the iliad Group, that rents virtual machines, GPUs, storage and managed services.

Scaleway Instances are KVM virtual machines you pick from fixed shapes, from small shared-vCPU Development and Basic Instances to dedicated-vCPU and Arm64 types, running Linux or Windows Server. You get root, SSH, a public IP and security groups, and you install and run your own software next to Scaleway products like managed databases, Kubernetes, object storage, Serverless Jobs and GPU Instances. It is general cloud infrastructure, not an agent sandbox, so this page compares the two where they overlap: giving an agent its own Linux machine.

Which use cases each one fits

For each use case, Battleships prices every sandbox provider on the same workload and sorts them from the cheapest monthly bill up. “29th cheapest of 365” means 28 of the other 364 providers run that workload, with every feature it needs, for less. “With limits” means the provider runs it but misses something, so it is listed after every full match. Open a use case for its whole ranking and what each provider misses.

Use case and its workloadGoogle Compute Engine: price rankScaleway: price rankboat: price rank
Software factory
100 at once · 4 vCPU / 8 GB · 2,200 sessions of 8 h a month
29th cheapest of 365
$2,607 a month
87th of 365, with limits
$322 a month
5th cheapest of 365
$634 a month
Always-on AI employees
50 machines 24/7 · 2 vCPU / 4 GB
45th cheapest of 361
$2,558 a month
158th of 361, with limits
$208 a month
6th cheapest of 361
$657 a month
Code interpreter bursts
100 at once · 1 vCPU / 2 GB · 200,000 sessions of 1 min a month
Per-second billing helps Compute Engine with short runs (#98 in our preset), while each Scaleway CPU Instance pays at least a full hour of uptime (#170). boat's start limits slow big bursts (#189).
98th of 336, with limits
$133 a month
170th of 336, with limits
$893 a month
189th of 336, with limits
$2,000 a month
Sandboxes for your users
300 at once · 2 vCPU / 4 GB · 30,000 sessions of 23 min a month
On both clouds you set up HTTPS access and the per-user lifecycle yourself, and 300 at once needs a quota request first. boat ranks #3 in our preset, ahead of Scaleway (#89) and Compute Engine (#97), with 100 to 2,000 sandboxes by plan and per-plan start limits.
97th of 349, with limits
$836 a month
89th of 349, with limits
$738 a month
3rd cheapest of 349
$209 a month
RL rollouts
1,000 at once · 2 vCPU / 4 GB · 1,000,000 sessions of 3 min a month
Many short rollouts favor Compute Engine's per-second billing, at $2,050 a month against $18,254 on Scaleway, where every run pays a full hour. All three are compromises here; boat pays for starts above its plan.
40th of 349, with limits
$2,050 a month
182nd of 349, with limits
$18,254 a month
39th of 349, with limits
$2,000 a month
Agent evals
250 at once · 4 vCPU / 8 GB · 15,000 sessions of 20 min a month
12th cheapest of 359
$228 a month
57th cheapest of 359
$776 a month
10th cheapest of 359
$200 a month
CI runners
40 at once · 8 vCPU / 16 GB · 6,000 sessions of 12 min a month
Builds that need /dev/kvm take extra work on both: on Compute Engine you enable nested virtualization per VM on an Intel family or N4D, and Scaleway does not document it. In our CI preset, boat ranks #2, Compute Engine #39 and Scaleway #89.
39th cheapest of 374
$359 a month
89th cheapest of 374
$777 a month
2nd cheapest of 374
$86 a month
24/7 dev boxes
1 machine 24/7 · 4 vCPU / 8 GB
A 24/7 dev box suits both clouds: a Scaleway DEV1-L comes to $41 a month (#32 in our preset) against $109 on Compute Engine (#84). boat ranks #19 at $26 with auto-stop off, and saves the disk every minute; its IP changes at each resume.
84th cheapest of 371
$109 a month
32nd cheapest of 371
$41 a month
19th cheapest of 371
$26 a month
Computer-use agents
50 at once · 4 vCPU / 8 GB · 20,000 sessions of 15 min a month
On both clouds you set it up: install a desktop and a remote-desktop server on a Linux VM, or use a Windows Server Instance over RDP on Scaleway. boat streams a 1920x1080, 60 fps desktop with Chrome and computer tools.
49th cheapest of 367
$720 a month
139th of 367, with limits
$168 a month
7th cheapest of 367
$200 a month
Browser-use agents
50 at once · 2 vCPU / 4 GB · 30,000 sessions of 10 min a month
On both clouds you install and wire the browser yourself. boat ranks #7 in our preset with Chrome and browser control ready in each sandbox.
86th of 356, with limits
$378 a month
238th of 356, with limits
$126 a month
7th cheapest of 356
$200 a month
Scraping & bots (startup scale)
50 at once · 1 vCPU / 2 GB · 100,000 sessions of 3 min a month
99th cheapest of 343
$256 a month
223rd of 343, with limits
$13 a month
6th cheapest of 343
$90 a month
Mass scraping & bots (undetectable)
1,000 at once · 1 vCPU / 2 GB · 3,000,000 sessions of 2 min a month
101st of 327, with limits
$2,644 a month
37th of 327, with limits
$253 a month
73rd of 327, with limits
$1,800 a month
Parallel red teaming
200 at once · 2 vCPU / 4 GB · 5,000 sessions of 30 min a month
21st cheapest of 351
$187 a month
158th of 351, with limits
$29 a month
3rd cheapest of 351
$200 a month
GPU eval jobs
10 at once · 8 vCPU / 32 GB · 500 sessions of 30 min a month
Both clouds rent Nvidia GPUs: Compute Engine up to B200 and GB300, and Scaleway B300, H100, L4 and L40S by the minute. boat has no GPU on any size.
Not ranked: misses a requirementNot ranked: misses a requirementNot ranked: misses a requirement
macOS agents
5 at once · 4 vCPU / 8 GB · 1,000 sessions of 20 min a month
Not ranked: misses a requirementNot ranked: misses a requirementNot ranked: misses a requirement

Key features

FeatureGoogle Compute EngineScalewayboat
What you getA general cloud VM from a global hyperscaler; you set up everything elseA general cloud VM in a European cloud; you set up everything elseAn agent sandbox: a full VM with agents, desktop, URLs and snapshots set up
IsolationEvery instance is a full KVM-based VM with its own kernelEvery Instance is a full KVM VM with its own kernelEvery sandbox is a full VM with its own kernel, on a physical host dedicated to boat; vCPUs are shared, not dedicated cores
Regions43 regions and 130 zones worldwide, including many in the EUParis, Amsterdam, Warsaw and Milan; you pick the zoneGermany, Finland and France; you cannot pick a region, and there is no US or Asia region
Billing unitPer second after a 1 minute minimumPer hour of uptime, including startup and standbyPer second, from a monthly plan ($20 minimum) whose price comes back as VM time for that month
Docker and nested virtualizationDocker, root and systemd; nested KVM once you enable it per VM on Intel families or N4D, not on E2, Arm or other AMD VMsRoot and Docker on Linux images; nested virtualization is not documentedDocker, root, systemd and /dev/kvm on every sandbox
StoppingA stopped VM stops CPU and memory charges; disks and external IPs keep billing, and the disk stays with no time limitStandby bills as running, and a shutdown from inside the machine puts it in standby; a console or API power-off stops compute, and volumes and IPs keep billingNo compute charge while stopped; disk kept at least 14 days; saved snapshots above the included count and the credentials proxy still charge
Resume with memorySuspend keeps a VM's memory for up to 60 days, and memory bills while suspendedStandby keeps the Instance running and billed; a power-off reboots from diskResume reboots from disk (no memory snapshots)
Snapshots and copiesSnapshots on demand or on a schedule you set; copies through a machine image and a new VMImages and volume snapshots on demand; scheduled snapshots through a Serverless Job you set up (official tutorial); copies are new Instances booted from an imageDisk snapshot every minute and at each stop; fork a running sandbox in one call, from its last 1-minute snapshot
Reaching the machineExternal IPv4 (static on request) and any port through firewall rules; for HTTPS you set up a load balancer with a Google-managed certificate, or a web serverFixed public IPv4 and IPv6 on any port; HTTPS needs your own proxy or a paid Scaleway Load Balancer with Let's Encrypt certificates, plus DNSUp to 50 HTTPS URLs, token-gated by default; public IPv6; no dedicated IPv4, and the IP changes at each resume
Coding agentsYou set it up: install the agent CLIs on the VM and drive them over SSH or startup scriptsYou set it up: install the agents on the Instance and drive them over SSHSeven harnesses preinstalled (Claude Code, Codex, OpenCode and more), with a prompt, steer and events API
Keeping keys out of the machineSecret Manager and service account tokens deliver secrets into the VM; for a key-injecting proxy you run your ownSecret Manager stores secrets that the machine fetches, so they end up on it; for a proxy you run your ownOptional paid credentials proxy ($1 per day on, plus $1 per 1,000 requests); other secrets still enter the sandbox
Desktop and computer useYou set it up: install a desktop such as Xfce and Chrome Remote Desktop (Google guide), then add your own click and screenshot toolsYou set it up: install a desktop and browser on a Linux Instance, or use a Windows Server Instance over RDP1920x1080, 60 fps desktop with Chrome, computer tools and an embeddable stream
Outbound network rulesVPC egress firewall rules by IP and port; domain rules on Cloud NGFW Standard, a paid tierSecurity groups with outbound rules by protocol, port and IP range; SMTP blocked by defaultOpen internet on every plan; no provider-side allowlist
How many at oncePer-region CPU quotas that vary by project; free trial accounts cannot request moreQuotas per shape, for example 10 DEV1-L after identity checks; more through support100 to 2,000 sandboxes at once by plan, with included starts per minute, hour and day
GPUs, Windows and ArmNvidia GPUs from T4 and L4 up to H100, H200, B200, GB200, GB300 and Vera Rubin; Windows Server; ArmGPU Instances with B300, H100, L4 and L40S by the minute; Windows Server; Arm64No GPU on any size; no Windows VM (Wine for Windows programs); x86_64 only
Every feature Battleships tracks (43)
FeatureGoogle Compute EngineScalewayboat
Full VM (own kernel)YesYesYes
Docker insideYesYesYes
Nested virtualizationNo documented wayNot publishedYes
RootYesYesYes
SystemdYesYesYes
Desktop GUIWith your own setupNo documented wayYes
BrowserWith your own setupNo documented wayYes
Computer-use APINo documented wayNo documented wayYes
SSHYesYesYes
Persistent diskYesYesYes
Pause/resumeYesNo documented wayYes
Automatic snapshotsNo documented wayWith your own setupYes
Fork/cloneYesYesYes
Live fork (no pause)YesYesYes
Memory forkNo documented wayNo documented wayNo documented way
Custom image (Docker or snapshot)YesYesYes
≥24 h sessionsYesYesYes
Idle auto-stopNo documented wayNo documented wayNo documented way
Wake on requestNo documented wayNo documented wayNo documented way
HTTPS preview URLsWith your own setupWith your own setupYes
Raw TCP inboundYesYesYes
Public IPv4YesYesNo documented way
Egress allowlistYesYesNo documented way
Secret proxyNo documented wayNo documented wayYes
Secret proxy for any APINo documented wayNo documented wayYes
Preinstalled agentsNo documented wayNo documented wayYes
Agent harness APINo documented wayNo documented wayYes
MCP serverNo documented wayNo documented wayNo documented way
GPUYesYesNo documented way
WindowsNo documented wayYesNo documented way
Arm64YesYesNo documented way
Self-hostingNo documented wayNo documented wayNo documented way
BYOCNo documented wayNo documented wayNo documented way
Open sourceNo documented wayNo documented wayNo documented way
Scoped API keysNot publishedNot publishedYes
Spend limitsNo documented wayNo documented wayYes
Usage per sandbox (API)YesNo documented wayYes
Audit logsYesYesNo documented way
SSOYesYesNo documented way
SOC 2YesNo documented wayNo documented way
HIPAAYesNo documented wayNo documented way
EU data residencyYesYesNo documented way
Zero data retentionNot publishedNot publishedYes

Performance

Google Compute EngineNot measured
ScalewayNot measured
boat28.06 runs/s

Neither Compute Engine nor Scaleway is in the public benchmarks we use, so we cannot say which is faster. Compute Engine speed depends on the machine family you pick, and Scaleway sells shared-vCPU Development types such as DEV1-L next to dedicated-vCPU types. For reference, boat did 28.06 runs per second on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB, second of 15 in StarSling's run of 3 October 2026 (run 37073680030). That boat sandbox ran on an AMD Ryzen 9 9950X host. boat also places some sandboxes on AMD Ryzen 7 9700X hosts, which this run did not measure.

V8 Web Tooling Benchmark at a 4 vCPU / 8 GB target, runs per second, higher is better. Source: StarSling's run of 3 October 2026 (run 37073680030), median of 5 sandboxes. Providers ran on different hardware in different regions. StarSling gives a shared rank when the run cannot separate two providers. Every boat sandbox in this run ran on an AMD Ryzen 9 9950X host. boat also places some sandboxes on AMD Ryzen 7 9700X hosts, which this run did not measure. Google Compute Engine and Scaleway are not in the public benchmarks we use, so we cannot say which is faster.

Pricing in five scenarios

Each bill is priced from the providers' public price lists by the open Battleships engine, with the cheapest plan that runs the scenario. Start limits, per-start fees and plan minimums are in the price, boat's included.

ScenarioGoogle Compute Engine, a monthScaleway, a monthboat, a month
One coding agent, workdays
4 vCPU / 8 GB, 50 GB disk, 8 hours a day, 22 days a month, files kept between days
$31 · Pay-as-you-go billing account (no plan fee)$14 · Usage account$20 · $20/mo
One agent that never stops
4 vCPU / 8 GB, 50 GB disk, running 24/7 all month
$64 · Pay-as-you-go billing account (no plan fee) · with limits$41 · Usage account$26 · $20/mo
50 agents in parallel
50 machines of 4 vCPU / 8 GB at once, 8-hour sessions, 1,100 sessions a month, Docker inside
$1,291 · Pay-as-you-go billing account (no plan fee)$537 · Usage account$317 · $20/mo × 3 seats
Sandboxes for your users
2 vCPU / 4 GB, 30,000 sessions of 10 minutes a month, up to 300 at once
$378 · Pay-as-you-go billing account (no plan fee)$734 · Usage account$200 · $100/mo × 2 seats
1,000 short jobs at once
2 vCPU / 4 GB, 1,000 at once, 3 minutes each, 200,000 jobs a month
$690 · Pay-as-you-go billing account (no plan fee)$4,892 · Usage account$8,800 · $2000/mo · with limits

Google Compute Engine: Compute Engine bills each VM per second after a 1 minute minimum at its list rate, here E2 sizes on demand in us-central1, with no plan fee. The E2 sizes we price have 16 GiB of RAM at 4 vCPU. Disks and external IPs bill on top, also while stopped, and the lowest always-on rate needs a 3-year commitment.

Scaleway: Each Scaleway Instance shape has an hourly price, billed per hour of uptime including startup and standby, so a short job still pays a full hour. The prices here include the disk but not a public IPv4 (about $4 a month), which bills on top and keeps billing while an Instance is powered off.

Scaleway: Plans cost $20, $100, $500 or $2,000 a month, and the whole plan price comes back as VM time for that month. A 4 vCPU / 8 GB sandbox costs $0.036 an hour, billed per second while it runs. Extra starts above the plan's included count cost $0.03 to $0.07. A stopped sandbox has no compute charge and keeps its disk for at least 14 days; saved snapshots above the included count and the credentials proxy still charge.

Limits in those scenarios

Compute Engine caps vCPUs per region with quotas that vary by project, and a free trial account cannot ask for more; one agent that never stops reaches $64 a month only with a 3-year commitment. Scaleway sets quotas per shape, such as 5 DEV1-L and 8 DEV1-M, rising to 10 and 15 after identity checks, so 50 agents and 300 user sandboxes need a support request. Hourly billing makes the 1,000-job burst cost $4,892 a month on Scaleway against $690 on Compute Engine. On boat that burst does not fit: the $2,000 plan includes 90 starts a minute, 600 an hour and 2,400 a day, extra starts cost $0.03 to $0.07, starts above 270 a minute fail with 429, and boat never queues a start.

ScenarioGoogle Compute EngineScalewayboat
One coding agent, workdaysNone for this scenarioNone for this scenarioNone for this scenario
One agent that never stops
  • 3-year commitment
None for this scenarioNone for this scenario
50 agents in parallelNone for this scenarioNone for this scenarioNone for this scenario
Sandboxes for your usersNone for this scenarioNone for this scenarioNone for this scenario
1,000 short jobs at onceNone for this scenarioNone for this scenario
  • 2,400 starts/day included, paid above that (needs ~6,667)
  • 600 starts/hour included, paid above that (needs ~1,000)
  • starts above 90/min are paid, and starts above 270/min fail with 429

Which one should you use?

Use Google Compute Engine when

you need regions outside Europe, per-second billing for short jobs, nested KVM on Intel machines, memory suspend, or a wide range of Nvidia GPUs.

  • Regions all over the world. Useful if your users or data sit outside Europe: Compute Engine runs in 43 regions and 130 zones.
  • Per-second billing for short jobs. Useful for bursts of short runs: Compute Engine bills per second after one minute, while a Scaleway CPU Instance bills a full hour.
  • Nested KVM and memory suspend. Useful if an agent needs nested VMs on Intel families, or must resume with its processes still running.

Use Scaleway when

you want a European provider, lower prices for machines that stay on, GPUs by the minute, or Windows and Arm64 servers in the EU.

  • Lower prices for machines that stay on. Useful for steady servers: a DEV1-L runs around the clock for $41 a month, less than an E2 VM, with no commitment.
  • A European cloud company. Useful if you want a European provider: Scaleway is part of the iliad Group and runs only in EU cities.
  • GPUs by the minute. Useful if agents train or serve models: Scaleway rents B300, H100, L4 and L40S GPU Instances by the minute.

Where boat fits

Compute Engine and Scaleway both rent general cloud VMs that start empty. If what you run is agents, boat gives each one a full VM in Germany, Finland or France with coding agents, a desktop, HTTPS URLs, snapshots every minute and no compute charge while stopped. Pick one of the two clouds when you need GPUs, Windows, a fixed IP or a wider cloud around the machine.

Agents you drive from outside, keys kept out

On both clouds you install, start and watch every agent yourself, and keys from Secret Manager end up on the VM, where any root process can read them, unless you run your own proxy. boat ships the agents with an API, and its optional paid credentials proxy keeps the real keys of supported AI providers outside the sandbox.

/dev/kvm, forks and no compute charge while stopped

Compute Engine's E2 VMs have no nested virtualization, and Scaleway does not document it. Both keep billing disks and IPs while stopped, and a copy means making an image and booting a new VM. A stopped boat sandbox has no compute charge and keeps its disk for at least 14 days; saved snapshots above the included count and the credentials proxy still charge.

Desktop, HTTPS URLs and fleets without a ticket

On both clouds you set up the desktop for computer use and the certificates and proxy for each HTTPS URL. 50 agents or 300 user sandboxes also need a quota request on both before launch. boat runs 100 to 2,000 sandboxes by plan, within per-plan start limits.

Use boat when what you run is agents: each gets a ready, full VM in Germany, Finland or France with coding agents driven from outside, an optional proxy for AI keys, /dev/kvm, a desktop, HTTPS URLs, snapshots every minute, forks and no compute charge while stopped. Check that its x86_64 Linux and start limits fit your work. Compare it one to one: boat vs Google Compute Engine and boat vs Scaleway.

boat's own limits
  • Plans start at $20 a month. The plan price comes back as VM time for that month. Unused time does not carry over. Source
  • Starts (create, fork, resume): the $20 plan includes 12 a minute, 60 an hour and 200 a day. The $2,000 plan includes 90 a minute, 600 an hour and 2,400 a day. Extra starts cost $0.03 to $0.07. A start fails with 429 above 3 times the included starts per minute, or 6 times per hour or per day. Source
  • A stopped sandbox has no compute charge. boat keeps its disk for at least 14 days. A saved snapshot keeps it longer; extra saved snapshots and the AI credentials proxy are paid. Source
  • Auto-stop defaults to 1 hour. You can turn it off on a paid plan. The trial allows 2 hours at most. Source
  • boat saves the disk every minute and at each stop. Memory, running programs and /tmp are not saved: a resume is like a reboot. A machine failure loses writes after the last good snapshot. Source
  • Machines run in Germany, Finland and France only. Snapshot copies can be stored outside the EU. Source
  • Public IPv6 and outbound IPv4. No dedicated IPv4. The IP changes at each resume. Source
  • Hosts are AMD Ryzen 7 9700X and Ryzen 9 9950X. vCPUs are shared. No GPU. You cannot boot your own Docker image. Source

Questions people ask

Is Google Compute Engine cheaper than Scaleway?

For machines that stay on, no: one coding agent on workdays comes to $14 a month on Scaleway against $31, and 50 agents in parallel to $537 against $1,291, before Scaleway's public IPv4. For short sessions Compute Engine costs less: 300 user sandboxes come to $378 against $734, and 1,000 short jobs to $690 against $4,892. On boat the first three come to $20, $317 and $200; boat's start limits do not fit the 1,000-job burst.

Which is faster, Compute Engine or Scaleway?

Neither is in the public benchmarks we use, so we cannot say. Speed on both depends on the machine type you pick. For reference, boat did 28.06 runs per second on the V8 Web Tooling Benchmark at 4 vCPU / 8 GB in StarSling's run of 3 October 2026, on an AMD Ryzen 9 9950X host.

Can Compute Engine and Scaleway run Docker and nested VMs?

Both run Docker, since each machine is a full VM with root. Compute Engine allows nested KVM once you enable it per VM on Intel families and N4D, not on E2, Arm or other AMD VMs. Scaleway's docs do not cover it, so you would test it on the shape you pick. On boat, /dev/kvm is on every sandbox.

Do I pay for a stopped VM?

On both clouds, disks and IPs keep billing while a machine is off. Scaleway also bills standby as a running Instance, and a shutdown from inside the machine puts it in standby. On boat, a stopped sandbox has no compute charge and keeps its disk for at least 14 days; saved snapshots above the included count and the credentials proxy still charge.

Where does my data run?

Compute Engine runs in 43 regions worldwide, including many in the EU, and you pick the region per VM. Scaleway runs in Paris, Amsterdam, Warsaw and Milan, and you pick the zone. boat runs sandboxes on machines in Germany, Finland and France, and you cannot pick a region. Snapshot copies can be stored outside the EU.

Do Compute Engine or Scaleway have an agent sandbox?

Not as VMs. Both sell general VMs, and you install agents yourself and drive them over SSH. boat is built for agents, with coding agents, a desktop and an optional credentials proxy in every sandbox.

Try boat

Install the CLI, then run boat new.

7-day free trial

More comparisons