Can a single switch really power a whole room of access points, cameras, and desk phones without shutting down mid-afternoon? Sometimes yes, sometimes no, and the difference almost always comes down to one number people ignore until something reboots: the PoE power budget. Getting that number right is less about fancy math and more about knowing which figures to trust and which marketing claims to discount.
This guide walks through the actual calculation, the standards behind it, and the mistakes that trip up even experienced installers.
What the PoE Power Budget Actually Means
Power over Ethernet lets a switch push both data and DC power down a single Ethernet cable. The switch acts as the PSE (power sourcing equipment), and the device on the other end, whether it is a camera, phone, or wireless access point, is the PD (powered device).
Every PoE switch has a total wattage it can deliver across all ports at once. That figure is the PoE budget. It is not the same as the switch’s power supply rating, and it is not the same as the sum of its per-port maximums. A 48-port switch might advertise 740 W of PoE budget while each port can theoretically deliver 30 W. Forty-eight ports at 30 W each would be 1,440 W, so clearly the switch cannot max out every port simultaneously. (Upcoming switch 2 games)
That gap is where confusion starts. The budget is a shared pool, and how a switch distributes it depends on the standard in play and the vendor’s firmware.
The Three PoE Standards That Set the Numbers
The wattage figures come from IEEE standards, and each generation raised the ceiling.
| Standard | Common Name | Max at PSE Port | Guaranteed at Device | Year |
|---|---|---|---|---|
| 802.3af | PoE (Type 1) | ~15.4 W | ~12.95 W | 2003 |
| 802.3at | PoE+ (Type 2) | ~30 W | ~25.5 W | 2009 |
| 802.3bt | PoE++ (Type 3) | ~60 W | ~51 W | 2018 |
| 802.3bt | PoE++ (Type 4) | ~90 W | ~71 W | 2018 |
Notice the two columns. The switch port delivers more than the device is guaranteed to receive. That difference is cable loss: some energy turns into heat as it travels down the copper. Longer runs and higher wattages widen the gap, which is why 802.3bt devices use all four wire pairs instead of two.
The practical takeaway: when you size a budget, use the PSE-side figure. The device’s draw is not what the switch has to supply.
Step One: Inventory Every Powered Device
Start with a plain list. For each device, note its PoE class and its actual draw in watts. Most manufacturers publish both on the datasheet.
The class matters because it determines how much the switch reserves. Under 802.3af and 802.3at, classes run from 0 to 4. Under 802.3bt, classes 5 through 8 were added for higher-power gear. A Class 4 device causes the switch to reserve roughly 30 W at the port, even if the device only pulls 12 W in normal operation.
This is the single most common misunderstanding in PoE planning. People add up the actual wattage printed on the device labels, compare it to the switch budget, and conclude they have plenty of headroom. Then the switch refuses to power the last few ports because it reserved the class maximum for every device.
Step Two: Decide How the Switch Allocates Power
Not all switches behave the same way. Two broad models exist.
Static or class-based allocation reserves the class maximum per port at the moment the device is detected. It is predictable and safe, but wasteful. A port with a 5 W device in Class 4 still locks up 30 W.
Negotiated allocation uses LLDP (Link Layer Discovery Protocol) or similar negotiation to learn the device’s actual draw and allocate only what is needed, plus a small buffer. This fits many more devices onto the same budget, but it depends on the device supporting negotiation and the switch firmware implementing it correctly.
Some switches also permit oversubscription, letting you connect more total device wattage than the budget allows on the assumption that not everything draws peak power at once. That works until a cold morning when every heated camera turns on its defroster simultaneously.
For planning purposes, assume class-based allocation unless you have confirmed otherwise in the switch documentation. It is the conservative choice, and conservative choices keep networks up.
Step Three: Run the Calculation
Here is the formula in plain terms:
Usable budget = switch PoE budget − reserved overhead − sum of per-port allocations
Then add a safety margin. Industry practice commonly lands on 10 to 20 percent, though this is convention rather than a standard requirement.
A worked example makes it concrete. Suppose a switch advertises a 370 W PoE budget and you plan to connect:
- 8 wireless access points, Class 4, drawing about 18 W each
- 12 IP cameras, Class 3, drawing about 9 W each
- 10 desk phones, Class 2, drawing about 5 W each
Under class-based allocation, the reservations look quite different from the actual draw:
| Device Type | Count | Class Max Reserved | Total Reserved | Actual Draw |
|---|---|---|---|---|
| Access points | 8 | 30 W | 240 W | 144 W |
| Cameras | 12 | 15.4 W | 185 W | 108 W |
| Phones | 10 | 7 W | 70 W | 50 W |
| Totals | 30 | 495 W | 302 W |
The actual draw of 302 W fits comfortably inside a 370 W budget. The class-based reservation of 495 W does not. A switch using strict reservation would refuse to power several ports, even though the real load is well under the limit.
This is exactly why the calculation has to match the switch’s behavior. If that same switch supports LLDP negotiation and allocation based on the negotiated draw, the deployment works fine. If it does not, the plan needs trimming or a bigger switch.
Step Four: Check the Power Supply Behind the PoE Figure
Here is a trap that catches people during procurement. A switch’s PoE budget often depends on which power supplies are installed. Many chassis and stackable models ship with a base PSU that supports a modest budget, and you only reach the headline figure by adding a second supply or upgrading to a higher-wattage module.
The marketing datasheet may quote the maximum achievable budget without making the PSU requirement obvious. Always check the footnote. If the switch needs two 1,100 W supplies to hit its stated PoE ceiling, that is a line item in the project cost, not a detail.
It also pays to confirm whether the quoted budget already accounts for the switch’s own consumption. Some vendors list a raw figure, others list a net figure after system overhead. The difference is usually tens of watts, which matters at the margins.
Common Mistakes That Sink PoE Deployments
A few errors show up again and again in real installations.
Adding up label wattage instead of class reservations. Already covered, and still the most frequent culprit.
Forgetting PoE on uplinks and infrastructure. Some deployments chain a small PoE-powered switch or extender, which itself consumes budget from the upstream port. That draw has to be counted.
Ignoring cable length and quality. Long runs on cheap cable increase loss. The device may still boot but behave erratically under load. For high-wattage 802.3bt devices, use cable rated for the job and keep runs within spec.
Skipping the margin. Networks grow. A budget calculated to 99 percent utilization leaves no room for the camera someone adds next quarter. The 10 to 20 percent buffer is not padding; it is operational breathing room.
Assuming all ports are equal. Some switches designate certain ports as high-power or reserve capacity for specific port groups. Read the port layout before assuming any port can host a 60 W device.
Where PoE Budgets Fit in Broader Hardware Planning
Power budgeting is one piece of a larger hardware decision. The same discipline applies when comparing platforms and planning upgrades. If a team is weighing a hybrid handheld against a traditional console, for example, the power and accessory ecosystem matters as much as the spec sheet, and comparisons like switch 2 vs steam deck show how peripheral and dock requirements shape the real cost. Network gear follows the same logic: the headline number rarely tells the whole story. (ITU)
For buyers mapping out a multi-year hardware plan, it helps to track how power standards evolve across product generations. Rumors about Nintendo switch 2 lite samsung 8lpu plus chip rumor 2026 circulate partly because efficiency gains change what a small device can do, and the same efficiency curve drives PoE hardware. Higher-efficiency chips mean more devices per watt, which quietly reshapes budget math.
Tools and Shortcuts That Actually Help
Spreadsheets beat mental arithmetic here. A simple sheet with columns for device name, class, class maximum, actual draw, and port number catches errors fast. Add a running total and a conditional format that turns red when reservations exceed 85 percent of the budget.
Vendor configurators exist too. Several switch makers publish online PoE calculators where you enter device counts and classes and get a compatibility verdict. They are useful sanity checks, though they tend to assume the vendor’s own allocation behavior, which is the point.
The most reliable shortcut is a power audit after deployment. Most managed switches report per-port power consumption in their interface. Compare the reported draw against your planning numbers a few weeks after go-live. If reality diverges sharply from the spreadsheet, either the class assumptions were wrong or the devices behave differently than the datasheet suggested. Either way, the audit catches it before a failure does.
Conclusion
Calculating a PoE power budget is not difficult once the pieces are clear. Inventory the devices and their classes. Understand whether the switch reserves class maximums or negotiates actual draw. Sum the reservations, add a margin, and confirm the power supplies can support the figure you are counting on.
The traps are almost all about mismatched assumptions: label wattage versus class reservation, headline budget versus PSU-dependent budget, static allocation versus negotiated allocation. Get those aligned and the math works out. Get them wrong and the network tells you, usually at the worst possible moment.
Frequently Asked Questions
What is the difference between PoE budget and PSU wattage?
The PSU wattage is how much power the switch can draw from the wall. The PoE budget is how much of that it can forward to connected devices. A switch with a 1,000 W supply might reserve several hundred watts for its own operation and deliver the rest as PoE. Always size against the PoE budget, not the supply rating.
Does a PoE device draw its class maximum all the time?
No. The class maximum is a reservation ceiling, not a constant draw. A Class 4 device might pull 8 W at idle and 25 W under load. The switch reserves based on the class unless it negotiates actual consumption, which is why class-based planning is conservative and negotiated allocation is efficient.
Can I exceed the PoE budget if devices rarely run at full power?
Some switches allow oversubscription, but it is a calculated risk. If enough devices hit peak draw at once, the switch will drop ports, typically the lowest-priority ones. For cameras with heaters or access points with bursty radios, that risk is real. Keep utilization under roughly 80 to 85 percent of budget for headroom.
How much does cable length reduce available power?
Cable loss grows with length and current. That is why the standard guarantees less power at the device than the switch delivers at the port. For standard runs under 100 meters, the difference is already baked into the class figures. Very long runs or thin-gauge cable can push loss beyond the standard assumption, so stick to spec-compliant cabling.
What happens if a device needs more power than its port can supply?
The switch will typically detect the mismatch during classification and refuse to power the port, or the device will fail to boot. Some switches log a power-denied event. The fix is a higher-class port, an 802.3bt-capable switch, or a local power supply for that device. (Power over Ethernet)
Do I need to recalculate the budget when adding devices?
Yes, every time. The budget is a shared pool, and each new device consumes reservation capacity. A quick spreadsheet update before purchase prevents the awkward discovery that the switch has no room left. This is also why the 10 to 20 percent margin exists: it absorbs small additions without a hardware refresh.
