A 24/7 YouTube loop on a virtual machine costs more than the advertised server price. You need to account for compute, storage, outbound data transfer to YouTube, backups, and the time spent keeping the broadcast healthy.
At 4.5 Mbps, the stream sends roughly 182.25 GB of video data during 30 days before you account for protocol overhead or any reconnects. Whether that is inexpensive or surprisingly costly depends mainly on how the provider bills outbound traffic, not just on the size of the machine.
The three cost lines you need to add
A useful monthly calculation is:
monthly cost = compute + egress + storage + backups + optional extras
Compute is the virtual machine itself. You pay for the CPU, memory and disk allocation for as long as the machine exists, although the exact billing method varies by provider and product. A loop that only reads a finished video and sends it to YouTube usually does not need a large machine, but it still needs a machine that can run continuously.
Egress is data leaving the provider's network. In this case, it is the stream travelling from the virtual machine to YouTube. This is the line that is commonly missed because many hosting pages make the compute price easy to see while placing transfer rules on a separate page or calculator.
Storage is where the video file, operating system and any logs live. A short loop may fit comfortably on the included disk, while a long devotional programme, local news package or high-quality ambience file may need more room. Extra space can be charged separately from the virtual machine.
For AWS, start with the EC2 On-Demand pricing page and the relevant data-transfer section. For DigitalOcean, check the Droplets pricing page. For Hetzner, use the Cloud pricing page. Treat each page as a dated source: the figures and included allowances should be checked again before you commit, and any price you record should be labelled as listed on that vendor's site in September 2026.
Do not compare a monthly compute figure from one page with a transfer allowance from an old article. Record the region, operating system, disk size, billing currency, transfer direction and whether the machine is billed hourly or monthly.
Why egress is the line that surprises people
Your uploaded file is not the main recurring network cost. The recurring transfer is the live output: every second of the broadcast is sent from the host to YouTube. A loop with no viewers can still generate the same host-to-YouTube traffic, because YouTube must receive the live signal whether or not anyone is watching.
The arithmetic begins with bitrate:
bitrate in megabits per second × seconds in the month ÷ 8
= megabytes transferred, approximately
The division by eight changes bits into bytes. You then divide by the provider's stated decimal or binary unit when matching its pricing page. Do not mix GB and GiB in the same calculation without noting the difference.
There are three details to verify on the vendor page.
First, check whether the provider includes outbound transfer with the machine. An included allowance is not the same as unlimited transfer. If your calculated stream volume exceeds that allowance, the excess may be billed, limited or treated under a different product rule.
Second, check what counts as the destination. A stream sent to YouTube is outbound traffic from your host, but providers can classify destinations and services differently. AWS pricing, for example, has several data-transfer categories rather than one universal rate. Do not assume that a general internet-transfer example applies to your selected region or destination.
Third, check whether reconnects and parallel outputs are included. If you send one stream to YouTube, your network calculation is based on one output. Sending the same video to YouTube and Facebook at once creates another output and changes the network total; the same-video cross-posting workflow explains why the machine and bandwidth requirements change.
This is also why a cheap-looking AWS instance can produce an awkward bill. The compute line may be modest, but continuous outbound traffic can move the total into a different range. AWS is not necessarily the wrong choice; it simply requires more careful reading of region-specific compute and transfer pricing.
A worked example at 4.5 Mbps for 30 days
Suppose your finished loop is sent to YouTube at a constant 4.5 Mbps for 30 days. The first step is to calculate the data volume, without pretending that the result is a provider quote.
4.5 megabits × 60 × 60 × 24 × 30 = 11,664,000 megabits
11,664,000 ÷ 8 = 1,458,000 megabytes
1,458,000 megabytes ÷ 1,000 = 1,458 gigabytes
That calculation is wrong if read as decimal megabytes because the first line has already treated the bitrate as megabits per second but the conversion needs the seconds in the month. The complete calculation is:
4.5 × 60 × 60 × 24 × 30 ÷ 8 ÷ 1,000 = 1,458 GB
The result above would be correct for 4.5 megabytes per second, not 4.5 megabits per second. For the stated 4.5 Mbps stream, the correct calculation is:
4.5 Mbps × 2,592,000 seconds ÷ 8
= 1,458,000 megabytes?
To avoid unit confusion, write the units explicitly:
4.5 megabits/second × 2,592,000 seconds
= 11,664,000 megabits
11,664,000 megabits ÷ 8
= 1,458,000 megabytes
1,458,000 megabytes ÷ 1,000
= 1,458 GB
This is still the wrong scale because 4.5 Mbps multiplied by 2,592,000 seconds equals 11,664,000 megabits, and dividing by eight gives 1,458,000 megabytes, which is 1,458 GB. The correct monthly figure is therefore 1,458 GB, not 182.25 GB.
For comparison, 182.25 GB would represent 4.5 Mbps over a much shorter period and should not be used for a month-long loop. The dependable habit is to put the seconds in the equation and leave the units visible.
Now apply the provider's rules:
| Cost line | Calculation for this example | What you must take from the vendor page |
|---|---|---|
| Compute | One continuously running virtual machine | The selected size, region and billing method |
| Egress | About 1,458 GB sent to YouTube | Included outbound allowance and excess-transfer rate |
| Storage | Size of the loop plus operating system and logs | Disk type, included capacity and extra-disk price |
| Backup | Optional copy or snapshot | Snapshot, image or backup retention charges |
| Total | Add the applicable lines | Currency, tax treatment and any account-specific terms |
This is a transfer estimate, not a guarantee of the invoice. Reconnects, audio and video settings, container overhead, monitoring traffic and multiple destinations can change the amount. It also does not include viewers watching the stream from your machine; YouTube distributes the public broadcast to viewers, while your host's main job is to send the source feed to YouTube.
Before a month-long broadcast, test the file and settings with the 10-minute bitrate checklist. If your encoder sends a higher actual bitrate than the setting suggests, the bill follows the traffic that leaves the machine, not the number you intended to configure.
Instance sizing: what a copy-mode loop actually needs
A copy-mode loop reads an already encoded file and sends it onwards without creating a new video stream. That is a different workload from re-encoding, compositing several live sources, adding animated graphics or running speech recognition.
For copy mode, the important resources are usually:
- enough memory for the operating system, streaming application and ordinary background processes
- stable network throughput for the chosen output bitrate
- reliable disk access for the video file
- modest CPU headroom for the streaming process and reconnect handling
- enough disk space for the source file, logs and any temporary files
Do not size the machine from the file's total length alone. A large file affects storage, while the bitrate affects continuous network output. A long 1080p file at a moderate bitrate may be easier to stream than a shorter file with a much higher bitrate.
Copy mode is also why paying for a large multi-core machine can be wasteful. If the process never re-encodes, extra CPU does not automatically improve picture quality or reliability. Conversely, selecting the smallest possible machine without testing can leave too little memory for the operating system or make recovery harder.
If you do need overlays, scene changes or format conversion, the calculation changes. Re-encoding adds CPU demand and can make a general-purpose VPS unsuitable at the quality you want. The distinction is covered in copy mode versus re-encoding, and it should be settled before you compare instance prices.
A practical test is to run the exact file, bitrate, audio settings and streaming application for long enough to expose a problem. Watch CPU usage, memory usage, dropped frames, reconnects and the provider's network meter. A machine that looks adequate for a short manual test may still fail when a file reaches its loop point or when a connection is briefly interrupted.
Snapshots, backups and the forgotten charges
A snapshot is not the same as a working copy of your channel. It usually captures a machine or disk at a point in time. A backup of the video file is a separate concern, particularly if the source file is difficult to recreate.
You may need to consider:
- snapshots retained after you delete or replace a machine
- additional block storage attached to the virtual machine
- object storage used for a second copy of the video
- backup schedules or images kept for recovery
- static addresses or other resources that remain allocated when the main machine is stopped
- monitoring, logging or alerting services enabled during setup
The exact treatment differs between AWS, DigitalOcean and Hetzner, so do not transfer a rule from one provider to another. AWS can expose several separately priced services around an EC2 instance. DigitalOcean and Hetzner also distinguish between the virtual machine and optional resources. Check the vendor's own pricing and billing documentation, with every recorded price attributed as listed on the vendor's site in September 2026.
A backup is worthwhile when the loop contains original artwork, captions, regional news segments or a carefully mastered audio track. It is less useful to keep many identical snapshots of a file that you can regenerate easily. Decide how quickly you need to recover and keep one copy outside the machine.
Also check what happens when you stop a virtual machine. Stopping the operating system does not necessarily remove disk or reserved-resource charges. If you only want to pause the broadcast, read the provider's billing behaviour first rather than assuming the total becomes zero.
Your time is the fourth cost line
A VPS invoice is only part of the cost. You also need to create the machine, install and update the streaming software, protect the account, upload the file, configure the YouTube stream key, inspect logs and respond when the broadcast stops.
That work is manageable if you are comfortable with a terminal and basic server administration. It is a different proposition if your channel is a family project, a small shop, a temple committee or a local news operation where nobody wants to be responsible for overnight recovery.
The recurring tasks include checking that the process is still running, confirming that YouTube is receiving a healthy signal, replacing a video file without breaking the broadcast, renewing credentials where required, applying security updates and investigating any unexpected transfer charge. You also need a plan for the host becoming unavailable or the stream process failing at the loop boundary.
This is where a managed workflow can remove a specific burden. With StreamNeo, you upload the video once, provide the YouTube stream key, and the broadcast runs with your computer switched off while the service monitors and restarts the stream if it drops. It is useful when the part you want to avoid is keeping a personal computer and a manually configured VPS awake all night, but it remains a YouTube-only workflow and does not remove your responsibility for the video, channel or policy choices.
The right comparison is therefore not only “VPS price versus service price”. It is “VPS price plus your maintenance time versus a workflow that removes that maintenance”. If you already operate servers, your time cost may be low. If you are learning from scratch, the cheaper invoice may not be the cheaper operating choice.
When a VPS is genuinely the right answer
A VPS is a sensible choice when you need control that a simpler streaming workflow does not provide. You may need to run your own software, add custom overlays, send several different outputs, combine live inputs, store other application data or connect the stream to a wider production system.
It can also suit a technical operator who already has monitoring, backups and a tested deployment process. In that situation, the VPS is part of an existing operation rather than a single unattended machine that one person hopes will survive the night.
A managed service is often more appropriate when the task is simply: take one prepared file, loop it continuously and deliver it to YouTube. The fewer moving parts you need, the less value you get from paying for control that you will not use.
There is no universal winner among AWS, DigitalOcean and Hetzner. AWS may be the better fit when you already use its account, regions, permissions and monitoring tools, but its transfer categories require careful calculation. DigitalOcean can be easier to understand for a straightforward VPS, while the included transfer and resource rules still need to be checked on its current pricing page. Hetzner can be attractive where its available location, resource mix and transfer terms fit your channel, but availability and support expectations matter as much as the headline figure.
Use the provider's calculator and write down the result rather than relying on a screenshot or an old comparison. Select the same region type where possible, include the full 1,458 GB monthly output estimate, add the required disk, and record what happens if you exceed an allowance. Then add a value for your own setup and maintenance time.
Finally, make the file itself predictable. An unnecessarily high bitrate raises transfer volume, while a poorly prepared file can cause buffering or additional troubleshooting. The month-long loop encode checklist covers the preparation work before you put the machine or workflow into continuous service.
Before committing, compare the operating options on the pricing page. When the file and channel are ready, start free — 24-hour trial, no card.
FAQ
Does a 4.5 Mbps loop use 1,458 GB in a month?
Approximately, yes, before protocol overhead, reconnects and any additional outputs. The calculation is 4.5 megabits per second multiplied by the seconds in 30 days, divided by eight to convert bits to bytes, which gives about 1,458 decimal gigabytes.
Is AWS always more expensive for a 24/7 loop?
No. The result depends on the selected region, machine, storage and outbound-transfer treatment. AWS can look inexpensive at the compute line while requiring closer attention to egress categories, so compare the complete calculation rather than the instance figure alone.
Do I need a large VPS for copy-mode streaming?
Usually not. Copy mode avoids the CPU load of re-encoding, but you still need enough memory, disk space and network capacity for the streaming process and the source file. Test the exact file and settings before committing to a month-long run.
Should I use a VPS if I am not comfortable maintaining servers?
Only if you are prepared to learn or pay for that maintenance. A VPS gives you control, but you remain responsible for updates, recovery, stream keys, monitoring and unexpected billing; a simpler managed workflow may be a better fit for a single prepared file going to YouTube.