Video conferencing cost: build, partner or SaaS, and what it depends on
If you are comparing per-host meeting subscriptions with running Jitsi or BigBlueButton yourself, the video conferencing cost turns on three things: peak concurrency, recording volume and who operates the stack. License fees are not the deciding line item, because both platforms are open source and carry no per-seat charge. Zoom's pricing page, as read in October 2026, lists a free Basic plan with 40-minute meetings for up to 100 participants, and paid Pro and Business plans billed per user per month with 30-hour meetings, so a subscription cost scales with headcount and hosts. A self-hosted cost scales with the number of simultaneous meetings, the media bandwidth they consume and the hours of operations labor you carry. Three options are on the table: build it in-house, partner with an implementation firm to build and then run it, or buy a hosted service such as the Jitsi as a Service offering from 8x8 or a vendor subscription.
Build costs are mostly one-time engineering plus recurring operations. Servers and bandwidth must be sized to the busiest hour, not the average day. The BigBlueButton 4.0 install documentation calls for at least 16 GB of RAM with swap, 8 CPU cores and 250 Mbit/s symmetrical bandwidth per server, plus 500 GB of disk for recordings (50 GB without recording). Jitsi's scalable setup guide suggests a separate central server and one videobridge per capacity unit, which you add as load grows. Recording storage grows with every recorded hour and needs its own retention rule, and TURN relays for users behind restrictive firewalls consume bandwidth of their own. Ask any partner to show a sizing model for your peak concurrency, your recording hours and your TURN share, and to state which of these recur monthly and which are one-time.
Partner and SaaS options move cost between fixed and variable lines. A partner quote follows from written scope, so ask for separate prices for initial build, hardening, load testing, monthly support, upgrades and any retainer, and confirm who holds the infrastructure contracts. DeployOpen does not publish prices. For a hosted service, confirm the per-host, per-minute or per-recording terms, where media and recordings are stored, and what it costs to leave. Whichever route you take, have finance and the platform team agree a single model of peak meetings, recording hours and staff time before comparing quotes, so the numbers describe the same workload.
Cost components for in-house, partner and SaaS video conferencing
This table compares who pays for each cost component under in-house, partner and SaaS video conferencing, and whether each cost is one-time or recurring. Read it as a checklist for requesting quotes: the hardware references are minimums and starting points, not sizing for your peak load. The main takeaway is that self-hosting shifts cost from per-host subscriptions to infrastructure, operations and upgrade effort that you must scope explicitly.
| Cost component | In-house build | Partner-delivered or SaaS |
|---|---|---|
| Media servers (Jitsi videobridges or BigBlueButton servers) | Your team pays for hardware or cloud instances. BigBlueButton 4.0 lists at least 8 CPU cores, 16 GB RAM with swap. Jitsi scalable setup suggests 4 or 8 CPU with 8 GB RAM per videobridge. Recurring. | Partner quotes hardware or cloud instances as part of scope. SaaS vendor carries media capacity inside per-host fees. Recurring when hosted. |
| Central services (Jitsi web, Prosody, Jicofo; BigBlueButton core) | Central server sizing starts at 4 CPU and 8 GB RAM for Jitsi. Your team pays for the machine and its administration. Recurring. | Included in partner service fee or SaaS subscription. Confirm in writing which components the quote covers. |
| Bandwidth | Your team pays data transfer from your provider. BigBlueButton 4.0 lists 250 Mbit/s symmetrical per server. Recurring, grows with concurrent participants. | Usually included in subscription tiers. Ask whether overage or regional relay charges apply. |
| Recording and storage | BigBlueButton 4.0 lists 500 GB disk with recording (50 GB without). Storage and backup are recurring. Jibri setup is one-time configuration effort. | Vendor may charge for cloud recording storage or retention. Ask for retention limits and export terms. |
| TURN servers | Your team pays for a relay host and its bandwidth. Recommended for users behind restrictive firewalls. Setup is one-time, operation recurs. | Typically included in SaaS. Partner quotes it as a separate line item. Confirm the relay location. |
| Load balancing and Scalelite | Your team pays for additional servers and configuration effort. Adding capacity repeats per server. | Included in SaaS scaling. Partner quote may include a one-time design and recurring tuning. |
| Monitoring, upgrades and support | Recurring staff time for patching, backups and incident response. Support contracts with BigBlueButton Inc. or 8x8 are optional and separately priced. | Partner retainers or support tiers cover recurring work. Confirm response expectations, upgrade windows and who tests before rollout. |
| Per-host subscription fees | None for open-source software. Infrastructure and staff replace these fees. Recurring costs for hosting continue. | Zoom publishes Pro and Business plans billed per user per month, with a free Basic plan (Zoom pricing page). Recurring per host, and scales with headcount. |
Why peak concurrency sets the server bill
Media servers are sized for simultaneous meetings, not for the total number of people in your directory. A company with 5,000 accounts may only need capacity for the handful of meetings that run at the same moment, and that peak is what sets the number of machines, the bandwidth commitment and the time spent on load testing. Before comparing any quote with a per-host subscription, ask the vendor or your own platform team to state the peak number of concurrent meetings, participants per meeting and share of video versus audio-only participation. Those three inputs drive the server bill far more than headcount does.
The published starting points are useful for scoping, but they are minimums, not guarantees. For BigBlueButton 4.0, the install documentation lists at least 16 GB of RAM with swap, 8 CPU cores with high single-thread performance, 500 GB of disk for recordings (50 GB without recording) and 250 Mbit/s symmetrical bandwidth on a single server. For Jitsi, the scalable setup guide suggests a 4 CPU, 8 GB machine for the central server and 4 or 8 CPU with 8 GB RAM for each videobridge, and the default install is already configured to add more videobridges horizontally. Treat these as baselines for a proof of concept. Confirm real capacity with load tests that mirror your meeting mix, recording settings and network paths, and budget for the second and third bridge or server if the test shows a shortfall.
BigBlueButton 4.0 cannot be reached by a direct upgrade from 2.6, 2.7 or 3.0, and the install documentation recommends a clean server installation. If you run an earlier branch, the migration is a separate project with its own planning, testing and downtime window, and it should appear as a line item in any build or partner quote. Ask whether the partner prices that migration, whether recordings and course data move with it, and how the 3.0 branch's maintenance releases will be handled during the cutover.
What moves the cost up or down
Most of the cost of a self-hosted meeting platform comes from the capacity you must buy for the busiest moment, plus the people who keep it running. Each driver below shifts that bill in a predictable direction, so measure these six inputs before you compare an open-source build with a partner engagement or a per-host subscription.
Peak concurrent participants and meetings
Compute and bandwidth are bought for the busiest hour, not the average day. Each Jitsi videobridge is a media router you add as load grows, and each BigBlueButton server is a fixed block of CPU, RAM and bandwidth, so the number of simultaneous meetings sets the server count. Measure concurrent participants and meetings at peak, including timetabled spikes such as the start of a term or an all-hands. Ask any partner for the sizing assumption behind each server count.
Recording and streaming volume
Recording is the cost line that grows with use and never stops. Storage for recordings is sized per hour recorded and multiplied by your retention period, and recording also adds CPU load to the server that hosts it. BigBlueButton 4.0 asks for 500 GB of disk with recording enabled and 50 GB without it, so turning recording on is a large, fixed step. Measure recorded hours per month, average file size and how long each class or meeting type must be kept.
Users behind restrictive firewalls
Participants who cannot reach media ports directly need a TURN relay, and that relay carries their audio and video as traffic through your network. The share of users on corporate, school or public networks that block UDP drives relay bandwidth and a second server to run and monitor. BigBlueButton's install docs recommend a TURN server for users behind restrictive firewalls. Measure the share of sessions that fall back to relay during a pilot, not the share of users who say they have problems.
Bandwidth symmetry
Video servers send as much as they receive, so upload capacity on the host often limits how many participants a server can serve. Symmetrical links cost more from most hosting providers and data centers than consumer-style asymmetric plans. BigBlueButton 4.0 requires 250 Mbit/s symmetrical bandwidth on its server. Measure peak upload and download separately per server, and confirm with your provider whether the price is metered by transfer or by committed capacity.
High availability and number of regions
A single server is the cheapest option and also the single point of failure for every meeting and class. Redundancy means a second set of servers, a load balancer, shared storage for recordings and a second region if you need continuity across an outage. Each added region multiplies the server count and adds replication traffic. Measure the outage you can tolerate for live classes and board calls, then price that requirement explicitly rather than assuming it is included.
Operator skill and retainer scope
Staff time is often the largest recurring cost and the one least visible in a comparison. Someone must patch the stack, rotate TLS certificates, watch capacity, handle upgrades around classes or meetings, and answer support tickets. Staff can be your own team, an implementation partner under a retainer, or a mix. Ask for the retainer scope in writing: which upgrades, monitoring hours, incident response and capacity reviews are included, and which are billed as extra work.
Build, partner or SaaS: which cost profile fits your organization
With an in-house build, your own staff select, deploy, secure and operate Jitsi Meet or BigBlueButton. The software itself is free under its open-source license, so the spending goes to people, hardware, bandwidth and time. Jitsi Meet is licensed under Apache-2.0 and BigBlueButton under LGPL-3.0, which means no license fee, but the work of running a media platform does not disappear.
- Staff time: engineers learn the component stack, configure authentication, handle certificates and maintain upgrades. This is usually the largest recurring cost and is hard to see in a license comparison.
- Hardware and hosting: BigBlueButton 4.0 is documented with at least 16 GB of RAM, 8 CPU cores, 500 GB of disk for recordings and 250 Mbit/s symmetrical bandwidth per server. Jitsi scales by adding videobridges, so capacity is planned per node.
- Bandwidth: media traffic is the cost that grows with concurrency and recording, and it is billed by your data center or cloud provider.
- Upgrades and incidents: upgrade paths differ by release, and BigBlueButton 4.0 does not support direct upgrades from 2.6, 2.7 or 3.0, so plan a clean installation.
Ask your team to estimate hours per month for patching, monitoring and on-call coverage before comparing this option with the others.
An implementation partner runs the work for you under a written scope, then either hands over the platform or keeps it supported under a retainer. You pay for the engagement in defined phases, and the infrastructure you keep running stays in your own cloud account, data center or isolated network. DeployOpen works this way: it plans, implements, hardens, documents and supports self-hosted platforms inside the client's infrastructure, and most engagements are fixed-scope.
- Scoping and design: a one-time cost covering meeting sizes, peak concurrency, authentication, recording and network paths such as TURN.
- Deployment and hardening: a one-time cost for installation, certificates, firewall rules, authentication setup and testing.
- Recurring infrastructure: servers, storage and bandwidth continue to be paid to your providers, and the partner cost does not replace them.
- Retainer support: an optional recurring fee for upgrades, monitoring, backups and incident response, sized to your availability needs.
- Documentation and handover: a fixed deliverable that reduces the internal staff time needed later.
DeployOpen does not publish a price list, so request a written quote that separates one-time work from retainer items and lists which infrastructure costs you will carry directly.
Commercial SaaS moves hosting and operations to the vendor and charges a subscription. Zoom's plan structure, as read in October 2026, has a free Basic plan with 40-minute meetings for up to 100 participants, Pro and Business plans billed per user per month with 30-hour meetings for 100 and 300 participants, and Enterprise by quote. Read the current per-user price from the vendor's pricing page, because it changes and this page does not state it.
- Per-host subscriptions: cost scales with the number of licensed hosts, not with concurrent meetings, so a small team with many meetings may pay less than a large team with few.
- Meeting limits: plan limits on meeting length and participant count set the features you can use without upgrading.
- Enterprise terms: contract, data processing and retention terms are negotiated by quote, so ask what recording and data location options are included.
- Hidden costs: integrations with your identity provider, legal hold, and recording storage in your own systems may need separate work.
- Data location: media relay, recording and support access sit with the vendor, which may conflict with sovereignty or sector rules.
Before deciding, compare the vendor's Enterprise quote with a self-hosted scope, including the data location and retention terms you need.
How to build your own video conferencing estimate
This sequence turns a meeting demand forecast into a line-by-line estimate for self-hosted Jitsi or BigBlueButton, so you can set the build cost against a per-host subscription for the same headcount. Each step produces a figure or check that feeds the final comparison.
Define peak concurrency and meeting shape
Start with the calendar, not the average. Count the maximum number of meetings running at the same time, the participants in the largest of them, and how many are external guests. Ask business owners for peak windows such as board days, term start or all-hands events, and record the typical meeting length and the share of meetings that need recording. The output is one table with peak concurrent meetings, peak participants per meeting and recorded hours per month. Without this table, every later figure is a guess.
Choose the platform and map server roles
Select Jitsi Meet for ad hoc, browser-first meetings or BigBlueButton for classes and structured sessions with whiteboards and breakout rooms. Map each role to a server: for Jitsi, the web server, Prosody, Jicofo and one or more videobridges, plus jibri if you record; for BigBlueButton, one or more full servers, with Scalelite if you pool them. Confirm which roles run on shared hosts and which need their own machines. The check is a role diagram that every server in the bill of materials maps to.
Size media and central nodes from the minimums
Use the vendor's published minimums as the floor for each node. The Jitsi scalable setup guide suggests a 4 CPU, 8 GB machine for the central server and 4 or 8 CPU with 8 GB RAM for each videobridge. A BigBlueButton 4.0 server needs at least 16 GB of RAM with swap, 8 CPU cores with high single-thread performance, and 500 GB of disk for recordings (50 GB without recording). Have your infrastructure team confirm the number of nodes needed for the peak load from step one, then price them on your cloud or hardware quote. The output is a node count per role.
Add peak bandwidth and TURN headroom
Multiply peak participants by the expected upload and download rate per stream, then add capacity for a TURN relay for users behind restrictive firewalls. BigBlueButton 4.0 lists 250 Mbit/s symmetrical bandwidth per server as a minimum, and Jitsi needs port 10000/UDP plus 443/TCP for fallback. Relayed traffic costs more than direct traffic on most cloud networks, so ask your provider for egress pricing per gigabyte and the share of users you expect to relay. The output is a monthly transfer figure and a bandwidth line item.
Price recording storage from hours and retention
Convert the recorded hours from step one into storage, using the retention period set by your records or compliance owner. BigBlueButton's install guide recommends 500 GB of disk for recordings on a 4.0 server, which gives a starting reference. Add the cost of backups, encryption at rest and any archive tier, and note who deletes recordings at the end of retention. The output is storage in gigabytes per year and a cost line for each storage tier, tied to a written retention rule.
Add load balancing, monitoring and TLS
Budget the components that keep meetings working under load: a load balancer or Scalelite for pooling BigBlueButton servers, monitoring for media node CPU, bandwidth and concurrent meetings, and certificates from a valid domain. Jitsi's handbook says self-signed certificates cause problems for working audio and video, especially on mobile, so include a certificate service or internal CA. The output is a list of these components with a one-time setup cost and a recurring run cost for each.
Add staff or retainer hours for operations
Estimate the people who patch, upgrade and support the platform. Separate the one-time installation and hardening from the recurring work of applying releases, testing upgrades, monitoring alerts and answering user issues. Decide whether an internal engineer, a partner retainer or a mix will cover this, and get the hours and rates in writing. Ask any implementation partner to price installation as a fixed-scope project and ongoing support as a separate retainer, so the recurring line is visible. The output is a yearly operations budget with named responsibilities.
Compare the annual total with per-host subscriptions
Add the one-time build costs spread over the planned service life to the yearly running costs, then divide by the number of licensed hosts you would need under a subscription. Check the current per-user, per-month price on the vendor's pricing page on the day you compare, because plan limits on meeting length and participants change those totals. Test the result against peak load, not headcount alone, since a per-host plan does not reflect your concurrency. The output is one side-by-side table with the assumptions written beside each figure.
Video conferencing cost questions buyers ask
How much does self-hosted video conferencing cost compared with per-host subscriptions?
Self-hosting replaces a per-user fee with infrastructure, staff time and support costs, so the comparison depends on how many people host meetings, how many run at the same time and how long you keep recordings. Zoom's pricing page, as read in October 2026, lists a free Basic plan with 40-minute meetings for up to 100 participants, and Pro and Business plans billed per user per month with 30-hour meetings for 100 and 300 participants. Build the comparison from your own peak concurrency, not from headcount alone, because a self-hosted server is sized for the simultaneous load. Ask a partner for a cost model that sets server, bandwidth, storage and labor lines beside your current subscription count.
What is one-time and what recurs in a self-hosted video deployment?
One-time costs usually cover design, installation, integration with identity and your LMS or chat tools, hardening, load testing and staff training. Recurring costs cover the servers and bandwidth that run every day, storage for recordings that accumulates, TURN relay capacity, monitoring, security patching, upgrades and support. Jitsi Meet and BigBlueButton are open source, so the software itself carries no per-user license fee, but the BigBlueButton 4.0 install docs list minimum server resources, including at least 16 GB of RAM, 8 CPU cores and 500 GB of disk for recordings, and those machines are paid for every month. Confirm with any vendor or partner which items are fixed and which scale with usage.
How long does a self-hosted video deployment take and what does it need from our team?
Duration depends on scope: a single Jitsi Meet or BigBlueButton server for a small group is a much smaller job than a multi-region deployment with SSO, recording, TURN and load balancing. Most engagements are fixed-scope, so the timeline is agreed in writing before work starts. Your team typically needs to provide a clean Linux host in the right region, a DNS name and a valid certificate, firewall changes for the required ports, an identity source to connect and named stakeholders for acceptance testing. For BigBlueButton 4.0, the install docs say direct upgrades from 2.6, 2.7 or 3.0 are not supported, so plan for a clean server installation. Ask the partner for the timeline and the client responsibilities in the written scope.
Can our internal team run self-hosted video conferencing without a partner?
Yes, if you have people who can operate Linux servers, reverse proxies, certificates, firewalls and container platforms, and who can monitor and patch them on a schedule. Jitsi's documentation covers Debian and Ubuntu quick starts, a Docker setup and a scalable setup guide, and BigBlueButton publishes its install and API docs. The harder parts are sizing for peak concurrency, configuring TURN for users behind restrictive firewalls, setting up JWT or LDAP authentication, and scheduling upgrades around live meetings or classes. Many teams keep operations in-house and buy a fixed-scope build or a retainer for the parts they lack. Compare your staff capacity with the scope before deciding.
What should we ask for in a video conferencing quote?
Ask for a written scope that lists every component and who owns it: web, prosody, Jicofo and videobridges for Jitsi, or the BigBlueButton servers and any Greenlight front end. Request sizing assumptions tied to your peak concurrency, participant counts and recording hours, plus the server, bandwidth and storage line items with their providers. Include TURN server design, load balancing, authentication setup, backup, monitoring, acceptance testing and documentation as separate lines. Ask which items are one-time, which recur, what a retainer covers, how upgrades are scheduled, and what happens when usage grows. DeployOpen does not publish a price list, and a quote follows from the written scope.
Does recording or TURN change the bill more than the meeting servers?
Often yes, and the answer depends on your usage. Recording adds storage that accumulates with every hour captured, plus retention and backup costs, so the BigBlueButton 4.0 docs recommend 500 GB of disk for recordings compared with 50 GB without recording. TURN relays media for users behind restrictive firewalls, which consumes bandwidth on the relay host, so a large share of remote participants can make it a major line item. The BigBlueButton install docs recommend a TURN server for restrictive networks, and Jitsi requires TCP fallback on 5349 for some clients. Ask the partner to model recording hours and the share of participants who will need relay, because those two numbers usually drive the largest variable cost.
How DeployOpen scopes and delivers a fixed-scope video conferencing implementation
DeployOpen plans, implements, integrates, hardens, documents and supports self-hosted open-source platforms inside your own cloud accounts, data centers or air-gapped networks. For build, partner or SaaS decisions, we start from written requirements on meeting sizes, peak concurrency, recording and bandwidth, then agree a fixed scope before any work begins.
How we scope
- We size meeting concurrency, participant counts and recording volume with you before naming any server or bandwidth figure.
- We document the cost components in the written scope: servers, bandwidth, TURN, recording storage and operations work.
- Engagements are fixed-scope, with deliverables and timelines agreed in writing before work starts, and a quote follows from that scope.
How we build
- We deploy Jitsi Meet or BigBlueButton in your infrastructure, with TURN, load balancing and recording configured to the sized capacity.
- We run security testing against the deployed platform, including authentication and exposed ports, before go-live.
- We set up monitoring that shows when concurrency approaches capacity, so added servers are a planned cost rather than a surprise.
How we hand over
- You receive architecture documentation, the cost model behind the sizing and runbooks for operations and recording storage.
- We document the controls and evidence the platform produces, so your compliance team can prepare for independent assessment.
- We walk your operations staff through upgrades, backups and capacity changes so the in-house cost of running the platform is clear.
How we support
- Ongoing support, upgrades and continuous testing are available as retainers, priced separately from the initial implementation.
- We can plan upgrades around scheduled classes or meetings, with the timing agreed in the written support scope.
- Ask us to quote a retainer alongside the build so the recurring cost of operations appears in the same comparison as SaaS subscriptions.
