A single live stream to 100 viewers at standard 1080p quality can already saturate a 1 Gbps connection. This surprises people who think of bandwidth in website terms, where even a busy page serves visitors in fractions of a second and moves on. Streaming does not work that way. Every viewer stays connected for the duration of the broadcast, pulling data continuously, and the total load is the bitrate multiplied by however many people are watching at once. A dedicated server for streaming has to be sized around that reality, not around typical web traffic.
This guide works through the actual bandwidth math behind hosting your own streaming platform, whether that is live video, video on demand, or both, and what else besides bandwidth genuinely matters once the numbers are right.
📖 The bandwidth fundamentals first
This guide builds directly on bandwidth concepts already covered in depth. Read What Is Network Bandwidth?, on the difference between bandwidth and throughput, and how pricing models actually work.
Why Streaming Is a Different Bandwidth Problem
Most websites move small amounts of data per visitor. A page loads, the browser fetches what it needs, and the connection goes quiet until the next click. The total bandwidth a busy website consumes is the sum of many brief, bursty transfers.
Streaming inverts this completely. A viewer who tunes in stays connected and pulling data for the entire length of the video, sometimes for hours. There is no idle moment between requests to average the load down. This is why the sizing question for a dedicated server built for streaming is not “how much traffic in total” but “how much sustained, simultaneous bandwidth at peak,” and the two numbers can differ enormously even for a modestly sized audience.
It also changes what kind of connection actually works. A metered connection, billed by the gigabyte, turns a successful stream into an unpredictable and rising cost precisely when things are going well. An unmetered connection, priced flat regardless of volume, is the only model that makes sustained, high-audience streaming financially predictable.
The Real Math of Concurrent Viewers
Here is where streaming bandwidth stops being abstract and becomes arithmetic. According to YouTube’s own live streaming documentation, a solid 1080p live stream at 30 frames per second calls for a bitrate of about 10 Mbps, rising to 12 Mbps at 60 frames per second. For 4K, YouTube recommends roughly 30 Mbps at 30 frames per second and 35 Mbps at 60.
Multiply that bitrate by the number of people watching simultaneously, and the total outbound bandwidth required becomes clear very quickly.
How many concurrent viewers saturate a 1 Gbps port
| Quality | Bitrate (YouTube spec) | Viewers to saturate 1Gbps |
|---|---|---|
| 1080p, 30fps | 10 Mbps | ~100 viewers |
| 1080p, 60fps | 12 Mbps | ~83 viewers |
| 4K, 30fps | 30 Mbps | ~33 viewers |
Bitrates per YouTube’s official live encoder documentation. A single 1Gbps port saturates faster than most people expect.
Table showing how many concurrent viewers saturate a 1 gigabit connection at different streaming qualities. Around 100 viewers at 1080p 30fps, 83 at 1080p 60fps, and only 33 at 4K 30fps.
Three digits’ worth of viewers is not a large audience. A moderately successful live event, a webinar, a product launch, or a niche community stream can reach these numbers without anything going viral. This is precisely why bandwidth headroom, not just processor or memory specifications, is the first thing to size correctly for this workload.
📖 Reaching a geographically wide audience
Viewers rarely sit in one place. Read Dedicated Server Hosting in Europe: The Amsterdam Hub, on serving a dispersed audience well from a single, well-connected location.
Why Metered Bandwidth Gets Expensive Fast at This Scale
The financial side of this math matters as much as the technical side. Cloud platforms typically charge per gigabyte for data leaving their network, a line item known as egress. At streaming volumes, this stops being a rounding error.
Take the 1080p example from the table above: 100 concurrent viewers at 10 Mbps for a single one-hour stream moves roughly 450 gigabytes of outbound data. Run that daily for a month, and the total climbs into tens of terabytes, all of it billed per gigabyte on a metered platform. A successful, growing stream produces a rising, unpredictable bill precisely when the content is working.
Unmetered bandwidth removes this variable entirely. The cost is fixed at the start of the month regardless of whether an audience of ten or a thousand tunes in, which turns a genuinely difficult forecasting problem into a flat, known number.
Where a Single Origin Server Reaches Its Limit
Honesty matters here, because a dedicated server with generous unmetered bandwidth is not infinite, and pretending otherwise would be the kind of overclaiming this guide should avoid.
A single server, however capable, has a ceiling set by its network port. Once an audience genuinely grows into the thousands of concurrent viewers, particularly if they are spread across distant regions, a single origin server starts to strain, both on raw bandwidth and on the latency viewers in far-away locations experience. At that scale, the answer is not a bigger single server. It is a content delivery network sitting in front of the origin, caching and distributing the stream from edge locations closer to viewers.
This is not a contradiction of everything above. It is a sequencing point. A well-specified dedicated server with unmetered bandwidth is the right foundation for the small-to-mid audience sizes most streaming projects actually reach, and it remains the origin even once a CDN is added on top for genuinely large scale. The dedicated server and the CDN are not competing solutions. One is the source, the other is the distribution layer that becomes necessary once the audience outgrows a single location.
What Else Matters Beyond Bandwidth
Bandwidth dominates the streaming conversation for good reason, but two other factors deserve honest mention.
Storage matters most for video on demand, where a growing library of pre-recorded content needs fast, reliable disks to serve without stutter under concurrent requests. NVMe storage handles this kind of high-throughput read load noticeably better than standard drives. Processing power matters when a server performs live transcoding, converting one incoming stream into multiple output qualities in real time, which is a genuinely CPU-intensive task distinct from simply passing bandwidth through. A server that only relays a single pre-encoded stream needs far less processing power than one transcoding several quality tiers simultaneously.
Neither of these replaces the bandwidth math above. They sit alongside it, relevant once the core bandwidth question is already answered correctly.
Bandwidth that does not meter your growth
Swify dedicated servers include 1Gbps unmetered bandwidth as standard, with upgrades up to 25Gbps for larger audiences. Enterprise SSD and NVMe storage, full root access, from a Netherlands data centre. From €120/month.
→ Explore Swify Dedicated ServersFrequently Asked Questions
Do I need a dedicated server for streaming, or is cloud hosting enough?
A dedicated server suits streaming particularly well because of unmetered bandwidth. Most cloud platforms bill for outbound data per gigabyte, and streaming generates outbound traffic continuously for as long as viewers stay connected, which makes costs rise unpredictably as an audience grows. A dedicated server with unmetered bandwidth keeps that cost fixed regardless of audience size.
Exclusive access to the hardware also avoids competing with other tenants for the same network port during a live event. Read What Is Network Bandwidth? for how unmetered and metered pricing actually differ.
How much bandwidth does live streaming actually need?
It depends on quality and concurrent audience size. According to YouTube’s own live streaming documentation, 1080p at 30fps needs roughly 10 Mbps, and 4K at 30fps needs about 30 Mbps. Multiplying that bitrate by the number of simultaneous viewers gives the total outbound bandwidth required, so a 1080p stream to 100 concurrent viewers already needs close to 1 Gbps of sustained bandwidth.
This is why bandwidth headroom, not just processor specifications, is the first thing to size for a streaming server. Read What Is Network Bandwidth? for the underlying concepts.
What happens if I exceed my bandwidth on a metered hosting plan?
On a metered plan, exceeding an allowance typically triggers per-gigabyte overage charges, or the provider throttles the connection, degrading stream quality for viewers at the worst possible moment, during a successful, high-traffic event. Streaming’s continuous, high-volume nature makes it especially prone to hitting these limits compared to typical website traffic.
Unmetered bandwidth avoids this entirely, since the cost stays fixed regardless of how much data a stream actually moves. Read Dedicated Server vs AWS Cost for how metered cloud pricing compares to a fixed dedicated server cost.
Do I still need a CDN if I have a dedicated server?
For small to mid-sized audiences, a well-specified dedicated server with unmetered bandwidth is enough on its own. A CDN becomes genuinely necessary once an audience grows into the thousands of concurrent viewers, particularly when they are spread across distant regions, because a single server’s network port and geographic location eventually become the bottleneck.
The two are not competing choices. The dedicated server remains the origin even once a CDN is added on top for larger scale. Read What Is a Content Delivery Network (CDN)? for how that layer works.
Does server location affect streaming quality?
Yes, particularly for live streaming, where a poorly connected server location adds latency and can increase buffering for distant viewers. A server hosted at a well-connected hub, with dense peering to many networks, reaches a geographically spread audience more consistently than the same hardware in a poorly connected location.
This matters most when an audience is not concentrated in one region. Read Dedicated Server Hosting in Europe: The Amsterdam Hub for how connectivity affects reach across a dispersed audience.
What server specs matter most for streaming besides bandwidth?
Storage and processing power both matter, in different ways. Fast NVMe storage matters most for video on demand, serving a library of pre-recorded content without stutter under concurrent requests. Processing power matters when a server performs live transcoding, converting one stream into multiple quality tiers in real time, which is considerably more CPU-intensive than simply relaying a single pre-encoded stream.
Neither replaces getting the bandwidth math right first, but both become relevant once that foundation is in place. Read NVMe and RAID Together: High IOPS Storage Guide for the storage side of high-throughput workloads.