The 60-second answer
Hosted PBX moves call control to a provider for a monthly fee, fully deductible in the year incurred. On-premise PBX means buying the system outright as Class 8 depreciable property at 20% declining balance. Hosted wins on cash flow and support; on-premise still wins on poor connectivity, high call volume, and strict internal control.
Start by rejecting the framing
Almost every comparison of these two architectures is published by someone who sells exactly one of them. The hosted vendors describe on-premise as a relic maintained by a retiring technician. The on-premise vendors describe hosted as renting something you could have owned. Both descriptions are marketing, and both are wrong often enough to be useless as a decision tool.
The useful framing is narrower. An on-premise PBX and a hosted PBX do the same job — they decide what happens to a call after it arrives — and they implement the same open signalling standard to do it [5]. Neither is the carrier; under the Telecommunications Act that role belongs to whoever owns or operates the transmission facility, which in most cases is a third party either way [4]. What differs is where the software runs, who is on the hook when it stops, and how the Canada Revenue Agency treats the money.
The tax difference, which is larger than most people expect
This is the part of the comparison that gets skipped, and it is frequently worth more than the sticker-price gap.
Buy a phone system and you have not bought an expense. You have bought property. Section 18(1)(b) of the Income Tax Act denies a deduction for “an outlay, loss or replacement of capital, a payment on account of capital or an allowance in respect of depreciation, obsolescence or depletion except as expressly permitted by this Part” [2]. What is expressly permitted is capital cost allowance, and the CRA is specific about where phone systems land: “electronic telephone equipment” sits in Class 8, at 20% declining balance [1].
Work that through on a $14,000 on-premise system. Ignoring the half-year rule for simplicity, the first year's deduction is roughly $2,800, not $14,000. The second is about $2,240. By year five you have deducted a little over half the purchase price, and the equipment is five years old. Meanwhile a hosted subscription is an ordinary current expense, deducted in the year incurred [3].
| Year | On-premise: cash out | On-premise: deduction (Class 8, 20% DB) | Hosted 25-ext: cash out | Hosted: deduction |
|---|---|---|---|---|
| 1 | $14,000 | ~$2,800 | $750 setup + $2,628 | $3,378 |
| 2 | $0 | ~$2,240 | $2,628 | $2,628 |
| 3 | $0 | ~$1,792 | $2,628 | $2,628 |
| 3-yr total | $14,000 | ~$6,832 | $8,634 | $8,634 |
Three-year cash out favours hosted by roughly $5,400 in this scenario, and hosted deducts every dollar it spends while the on-premise buyer has deducted under half. That is before any maintenance labour is counted on the on-premise side. If you are comparing a purchase made years ago against today's subscription, restate the old dollars using the Bank of Canada's CPI calculator first [6] — a $14,000 system bought in 2016 was a materially larger commitment than $14,000 sounds today.
None of this makes on-premise the wrong answer. It makes on-premise a capital answer, which is a different kind of decision than an operating one, and should be evaluated against your cost of capital rather than against a monthly fee.
Where the risk actually sits
| Failure | On-premise | Hosted |
|---|---|---|
| Hardware dies | Yours to replace, on your timeline and budget | Provider's problem, invisible to you |
| Internet outage at your site | Internal calling survives; outside calls stop unless copper remains | Fails over to mobile or voicemail-to-email |
| Power outage at your site | Down without a UPS and generator | Fails over; desk phones down, routing continues |
| The person who configured it leaves | Serious. This is the most common real-world failure | Provider retains the configuration |
| Security patching | Yours. An unpatched internet-facing PBX is a genuine liability | Provider's |
| Provider goes out of business | Not applicable | Real risk. Ask about configuration export before signing |
| You want an unusual call flow | Anything is possible if you can build it | Bounded by what the provider supports |
The row that decides more migrations than any other is the fourth one. An on-premise PBX configured by a capable person is an excellent system right up until that person retires, resigns, or stops returning calls. What remains is a working box nobody can safely change — and a business that cannot change its own after-hours greeting.
When on-premise is still the right call
Four situations, none of them nostalgic:
- The connectivity is not there. Rural and semi-rural Canadian sites with a single unreliable connection should not put call control on the far side of it. This is a real constraint across a great deal of the country, not an edge case.
- Call volume is high and sustained. Contact centres and dispatch operations with heavy concurrent call loads can reach a scale where owning the infrastructure is cheaper per call, and where control over media handling matters.
- Internal control requirements are strict. Some organisations need call recordings and voicemail to never leave a building they control. That is a legitimate requirement and it is satisfiable on-premise without argument.
- You already own a working system and the expertise to run it. Replacing a functioning PBX because it is unfashionable is the most expensive form of tidiness there is.
The honest recommendation is that this decision is made at a site visit, not in a spreadsheet. Someone should see the wiring closet, the internet service, the existing handsets and the copper before an architecture is chosen. Hybrid arrangements — copper terminating on-site through an FXO gateway with call control hosted, or a headquarters system with hosted extensions for remote staff — are frequently the correct answer and are almost never on the pricing page of a vendor that sells only one side.
When hosted is clearly better
- Under 100 extensions, which describes the overwhelming majority of Canadian establishments [7].
- Multiple sites or remote staff who need to be on one dial plan.
- No internal telephony expertise, and no desire to acquire any.
- An existing PBX that is unsupported, undocumented, or configured by someone unreachable.
- A preference for a predictable monthly number over an unpredictable repair bill.
What hosted costs, without a quote form
MapleReceptionist publishes flat bands rather than per-seat rates, in Canadian dollars, HST added at checkout by province: $99/month up to 10 extensions, $219 up to 25, $399 up to 50, $699 up to 100. Setup is $350 at the entry band, or $250 with a 12-month commitment; $750, $1,200 and $1,900 at the larger bands. Over 100 extensions is quoted individually.
Bands are ceilings, not allowances. Adding people mid-term does not change the invoice; you move to the next band at renewal, with the number known in advance. Included: an extension per person, Canadian-hosted voicemail with voicemail-to-email, auto-attendant and IVR design, business-hours and after-hours routing, SIP trunk or analogue FXO integration, provisioning of handsets you already own, and roughly 30 to 60 minutes of administration per month. Numbers and DIDs, SMS, and AI call handling are separate products — a hosted PBX that bundles them is not cheaper, it is a larger purchase with a smaller label.
Bottom line
Hosted PBX converts a capital decision into an operating one, moves the failure risk to someone whose job it is, and deducts fully in the year you spend it. On-premise PBX keeps control and the media path inside your building and can be cheaper at volume, at the cost of a depreciation schedule [1][2] and a dependency on someone who knows the system. Both are correct for real Canadian businesses in 2026. Anyone who tells you otherwise is describing their product line, not your situation.