MHOSTMHOST

VPS vs Cloud (AWS, Google Cloud): Which Is Cheaper for a Small Business

Short answer: for a website, an online store, a CRM or a handful of internal services, a VPS is usually 2–5 times cheaper than AWS or Google Cloud for the same resources, and you know the bill in advance. The cloud starts to pay off when your load swings sharply, you need ready-made managed services, or you have a DevOps engineer on the team.

Below we break down what a cloud bill is made of, where the unexpected costs hide, and how to tell what your business actually needs.

The key differences

Pricing model. A VPS has a fixed monthly price: CPU, RAM, disk, IP address and traffic are already included. The cloud is pay-as-you-go: you pay separately for server hours, gigabytes of storage, outbound traffic, IP addresses, snapshots and service requests.

Scaling. A VPS grows by switching to a bigger plan. In the cloud you can add servers automatically when load rises and remove them when it drops, but that has to be set up.

Complexity. A VPS is one server with root access that any developer can handle. AWS has more than 200 services, and even a simple website means dealing with VPC networks, security groups, IAM permissions and disk types.

Performance. Cheap cloud instances (such as the AWS t series or Google Cloud e2-micro) run on CPU credits: once they run out, the server slows down or starts costing more. On a VPS your allocated cores run constantly with no such limits.

Hidden cloud costs

The server price on a cloud provider's website is only part of the final bill. Here's what gets added on top:

  • Outbound traffic. On AWS the first 100 GB a month are free, then it's about $0.09 per GB. A site with images and video serving 1 TB a month pays around $80 extra for traffic alone. On a VPS that volume is usually included in the plan.
  • Public IP address. Since 2024 AWS charges $0.005 per hour for every IPv4 address — about $3.60 a month per address.
  • Disks and snapshots. Storage is billed separately from the server, fast high-IOPS disks cost much more than basic ones, and backup snapshots are charged per gigabyte too.
  • Managed services. An RDS database, a load balancer, a NAT gateway — each is billed by the hour and often costs more than the server itself.
  • Forgotten resources. A test server, an undeleted disk or an old snapshot keeps costing money until someone removes it.
  • Billing after the fact. The cloud bills you in dollars at the end of the month, and you only see the exact amount once it's charged.

A separate risk is an attack or a bug. If bots start mass-downloading files or a script gets stuck in a loop, the cloud happily scales up and sends a bill for hundreds or thousands of dollars. On a VPS the worst case is a slow server, while the price stays the same.

Cost example: an online store

The task: a server with 2 vCPU and 4 GB RAM, an 80 GB disk and about 500 GB of outbound traffic a month.

AWS (EC2 t3.medium, on-demand, US region):

  • server — about $30;
  • 80 GB gp3 disk — about $6.40;
  • public IPv4 — about $3.60;
  • traffic: 400 GB beyond the free 100 — about $36.

That's roughly $75 a month, not counting backups, a load balancer or a managed database. Paying a year upfront makes the server cheaper, but traffic and IP stay at the same price.

A VPS with the same resources: one fixed plan price that already includes the NVMe disk, IP address and traffic. On the market such a server costs several times less, and the price doesn't change from month to month.

Cloud prices are approximate and vary by region; for an exact estimate for your workload, use the AWS Pricing Calculator or the Google Cloud Pricing Calculator.

When to choose a VPS

  • A website, online store, blog or landing pages with predictable load.
  • A CRM, accounting software, n8n, Telegram bots, a VPN and other internal services.
  • You need to know your budget in advance, with no surprises at the end of the month.
  • There's no dedicated DevOps engineer on the team.
  • Lots of outbound traffic: images, video, downloadable files.
  • Your audience is in one region, and you need a server close to it.

When the cloud makes sense

  • Load spikes by multiples: TV ad traffic, ticketing services, seasonal sales.
  • You need ready-made services: managed databases, queues, machine learning, GPU servers for a couple of hours.
  • The project serves users on several continents from day one.
  • Partners or investors require specific certifications or infrastructure.
  • Your startup has cloud credits. But once they run out, you pay the bill yourself, and leaving the cloud gets harder.

Often the best option is a mix: the main site and services live on a VPS, and the cloud is used selectively for tasks a VPS can't handle.

FAQ

What about the AWS and Google Cloud free tier?

It's designed for trying things out: the servers are weak and limited in resources and time. It won't be enough for a production site, and any service outside the limits becomes paid right away.

Is the cloud more reliable than a VPS?

A single cloud server has roughly the same availability as a good VPS. The cloud only gives you fault tolerance if you set up several servers in different zones, and that multiplies the bill.

Can I move from the cloud to a VPS?

Yes, as long as the project isn't tied to the provider's proprietary services. A site with a database or an app in Docker can be moved in a few hours.

What should a startup choose?

Without cloud credits or a DevOps engineer — a VPS. It's cheaper and simpler, and you can move to the cloud once there's a real need for it.

← All articles