From PSTN to Voice over Cloud: Four Eras

Ask an Australian business what kind of phone system they have and the answer is usually a brand name, occasionally a technology, and almost never a generation. That is a problem, because the generation is the part that determines what happens next. A business on analogue copper lines, a business running an ISDN PBX in a comms cupboard, a business that bolted SIP trunks onto that same PBX in 2016, and a business on a cloud platform are four genuinely different situations with four different sets of options, four different risk profiles and four different bills. They are also, awkwardly, four situations that can all be described as having a phone system that works. This article lays out the four eras of business telephony — analogue POTS, digital ISDN with an on-premises PBX, VoIP, and Voice over Cloud — and what actually changed at each transition. It spends time on the distinction most buyers get wrong, which is that VoIP describes how voice travels and Voice over Cloud describes where the system lives, meaning they are separate decisions that get sold as one. And it finishes with a diagnostic you can run in five minutes to work out which era you are in, plus the honest arithmetic on when moving is worth it and when it is not.

Technology History · Migration · 2026

One Hundred and Forty Years in Four Steps

Every Australian business phone system belongs to one of four generations, and most owners could not tell you which. That matters, because the sensible next move is entirely different depending on where you are starting — and because the single most common mistake in this market is treating VoIP and cloud as the same decision when they are not even the same kind of thing.

📅 ⏱ 16 min read 🇦🇺 Australian owned, Australian hosted, Australian supported
TL;DR

Four eras, not two. Analogue POTS from the 1880s, where one line carried exactly one call and every feature was metered separately. Digital ISDN with an on-premises PBX from the late 1980s, where channels replaced lines, call control moved into your building, and capacity became a hard ceiling that required a site visit to raise. VoIP from the 2000s, where voice became data packets and SIP trunks replaced circuits — but the PBX frequently stayed exactly where it was, which made call quality your problem rather than the carrier's. And Voice over Cloud from 2017, where there is no box, capacity is elastic, any device is an extension and you pay per user per month. The distinction that matters: VoIP is a transport technology, Voice over Cloud is a delivery model. You can run VoIP with the system still in your cupboard, and thousands of Australian businesses do. Copper is retiring, which removes the option of doing nothing. Two paths remain — SIP trunk the existing PBX, or move to a platform and skip a hardware generation. Below roughly fifty seats, keeping the box alive is an expensive way to defer the decision. Above a hundred, owning the platform can start to win again.

Why the Generation Matters More Than the Brand

Technology histories are usually written for the pleasure of the writer. This one has a use: the era your business is in determines which options are actually available to you, and businesses routinely receive advice pitched at a generation they are not in.

A business on analogue lines being sold SIP trunks is being sold a bridge to a building it does not own. A business with a functioning ISDN PBX being told to rip it out this quarter is being sold urgency. A business already on SIP trunks being told it is "on the cloud" is being told something that is not true and will cost it during the next outage. Each of those conversations happens weekly in this market, and each is only detectable if you know which era you are in.

4
Distinct eras, all still in service in Australia today
1
Calls one analogue line can carry, ever
2017
When the Voice over Cloud model was named
2
Migration paths available once copper goes

Era One: Analogue POTS, 1880s to 2000s

The plain old telephone service. Voice travelled as an electrical signal down a pair of copper wires, and the fact that this worked at all for a century is one of the more impressive engineering achievements in the built environment.

CharacteristicWhat it meant in practice
Voice as an electrical signalNo conversion, no packets, no software. The medium was the message, literally.
One line carried exactly one callTwo simultaneous calls required two lines. Ten meant ten. Capacity was a physical property of the building.
Call control sat in the carrier exchangeYou owned a handset and rented a service. Everything intelligent happened somewhere else, operated by somebody else.
Features were metered and billed individuallyCall waiting was a line item. Call forwarding was a line item. Voicemail was a line item. This is where the habit of billing per feature was learned, and the industry has never entirely lost it.
The one genuine virtue, and it is a real one

An analogue line was powered from the exchange, which meant a corded handset kept working when the building lost power. No subsequent generation has replicated that for free, and any honest account of this history has to concede it. What every subsequent generation offers instead is redundancy by other means — a mobile app in a pocket, a diversion that fires when a site goes dark, a platform that reroutes to another network. Different mechanism, better outcome in most real failures, but not the same thing, and businesses that lived through a copper-era blackout are entitled to notice the difference.

Era Two: Digital ISDN and the On-Premises PBX

From the late 1980s to, in a surprising number of Australian comms cupboards, the 2020s. The Integrated Services Digital Network digitised the local loop, and the private branch exchange moved the intelligence into the building.

Two things changed, and both are still shaping how businesses think about phone systems decades later.

🔢

Channels replaced lines

Instead of one physical pair per conversation, a digital service carried a fixed number of channels. More efficient, more manageable, and still a fixed number — which is the seed of the next problem.

🏢

Call control moved on-site

The PBX in your building decided what rang where. For the first time a business could design its own call flow rather than order features from a carrier's list. This was genuine progress and it created genuine dependency.

🚧

Capacity became a hard ceiling

Your system handled exactly as many concurrent calls as you had provisioned channels for. Call eleven on a ten-channel service did not queue, degrade or cost extra — it failed. Businesses learned to size for their worst hour and pay for it all year.

🚚

Every change was a site visit

New starter, moved desk, changed hunt group, different after-hours message: a technician, a booking, an hourly rate and a wait. The cost of change was high enough that businesses stopped changing things, which is its own hidden cost.

The lasting damage of era two is cultural, not technical. A generation of Australian businesses learned that the phone system is a fixed asset you do not touch, that changes cost money and take a fortnight, and that capacity is something you buy once and live with. Those assumptions are wrong now and they are still the default posture in a lot of boardrooms — which is why "we could not change the call flow" is still offered as an explanation rather than treated as an outrage.

Era Three: VoIP, and the Mistake Everyone Makes

From the 2000s to right now, because this is where a very large number of Australian businesses actually sit. Voice became data packets. SIP trunks replaced digital circuits. The bill usually went down.

And here is the part that gets skipped: the PBX often stayed exactly where it was.

A business that put SIP trunks into an existing on-premises PBX in 2014 changed its transport and nothing else. The box in the cupboard still ran the call flow. Changes still needed somebody who understood that box. Capacity was now bounded by bandwidth rather than by channel count, which is more elastic but not elastic. And one entirely new burden arrived that had not existed in either previous era.

Quality became your problem

On analogue and ISDN, call quality was the carrier's responsibility and it was essentially a solved problem. Once voice became packets on a general-purpose network, quality became a function of your internet, your router, your upload headroom and your local network — and when calls broke up, the carrier was entitled to point out that their service was within specification. This is the origin of every jitter, latency and one-way-audio conversation that has happened in an Australian office since about 2008, and it is why upload capacity matters far more than the download figure businesses actually shop on.

Era three is genuinely better than era two. It is also frequently mislabelled, and the mislabelling is expensive. A business running SIP trunks into its own PBX has a single point of failure sitting in a cupboard in one building, on one power supply, on one internet connection. When somebody tells that business it is "on the cloud", the business stops planning for the failure it actually has.

Era Four: Voice over Cloud

From 2017, when the delivery model acquired a name distinct from the transport that carries it. The defining characteristic is simple: the call control software lives in the provider's data centre, and your phones and apps connect to it over the internet. There is no box.

What changedConsequence for a business
No box to ownNo hardware refresh cycle, no end-of-support cliff, no single cupboard whose flooding takes out the phones, no capital purchase to depreciate.
Capacity is elasticConcurrent call capacity scales on demand rather than being provisioned to a worst-case peak and paid for permanently. The eleventh call is not a failure.
Any device is an extensionDesk phone, laptop, mobile, browser — the same extension, the same number, the same call flow. Location stops being a property of the phone system.
Pay per user, per monthCost tracks headcount rather than infrastructure. Adding a person is a setting; removing one is a setting.
Automatic failoverAn outage at one site does not take the number down, because the number never lived at the site.
Self-service provisioningA new user takes minutes in a portal rather than a technician visit. The cost of change collapses, which changes what businesses are willing to try.
Continuous upgradesNew capability arrives without a scheduled outage or a version migration project. You do not run version 6.2 of anything.
Native AITranscription, summaries and AI agents are platform features rather than bolt-on products, because the audio is already in the platform. This is the capability gap era three cannot close.

That last row is the one that has moved fastest since 2017 and the one that will keep moving. An on-premises system can be given AI features only by exporting audio somewhere else, which means an integration, a second vendor and a data-handling question. A platform that already holds the audio can transcribe, summarise, score and answer calls as ordinary functions. The gap is not that cloud is newer. It is that the audio is in the right place.

Transport Versus Delivery: The Distinction That Costs Money

If you take one thing from this article, take this.

VoIP is a transport technology. Voice over Cloud is a delivery model. VoIP describes how the voice travels — as data packets rather than an electrical signal. Voice over Cloud describes where the system lives — in the provider's platform rather than in your building. They are different axes. You can have one without the other, and a great many Australian businesses have exactly one.

ConfigurationTransportWhere the system livesCommon descriptionAccurate?
ISDN PBXCircuitYour cupboard"Traditional phone system"Yes
SIP trunks into your own PBXPackets (VoIP)Your cupboard"We're on VoIP" / "we're on the cloud"First one yes, second one no
Hosted PBX / cloud platformPackets (VoIP)Provider's data centre"Cloud phone system"Yes
Voice over CloudPackets (VoIP)Provider's platform, elastic, app-first, AI-native"Cloud phone system"Yes, and the distinction from a lifted-and-shifted hosted PBX is real

The commercial consequence of conflating the two is that businesses in row two believe they have the resilience properties of rows three and four. They do not. The single point of failure is still in the cupboard, and the discovery usually happens during an event when nobody has time for a taxonomy lesson.

What Happened to Capacity at Each Step

Concurrent call capacity is the cleanest way to see the four eras, because it is the constraint that shaped every other decision.

EraConcurrent callsTo get more you...What over-provisioning cost
Analogue POTSOne per lineInstall another physical lineA permanent line rental for a peak that happens twice a year
Digital ISDNFixed channel countOrder more channels, wait, pay a site visitChannels billed monthly whether used or not
VoIPBounded by bandwidthBuy more upload headroomLess, but quality degrades before capacity does, which is worse
Voice over CloudElastic, scales on demandNothingNothing — you are not provisioning it

The pattern is a hundred and forty years of businesses paying to be ready for their busiest hour, and then one step where that stops being a thing you buy. For a business with genuinely spiky call volume — a clinic on Monday morning, a venue on Friday afternoon, an accountant in July — that single change is worth more than every feature in the platform.

Follow the Call Control

There is a neater way to tell the four eras apart than any specification sheet: ask where the decision about what rings where is actually made.

🏛️

Era one — the carrier exchange

Somebody else's building, somebody else's software, and you order features from a list. Zero control, zero maintenance.

🗄️

Era two — your cupboard

Your building, your box, your consultant. Full control, and full responsibility for a physical asset in one location.

🔌

Era three — still your cupboard

The transport modernised and the control did not move. This is the era most businesses misdescribe, and it carries era two's risk with era three's bill.

☁️

Era four — the provider's platform, controlled by you

The software runs somewhere resilient; the configuration is yours, in a portal, changeable in a minute. Control without custody — which is the actual innovation.

The Australian Context: Copper Is Retiring

None of the above would be urgent if the first two eras were stable. They are not. Australia's copper network is being progressively decommissioned, and PSTN and ISDN services are being retired with it.

That removes the option a lot of businesses were quietly exercising, which was to do nothing indefinitely. It is worth being precise about the pressure, because it is often overstated by people selling and understated by people procrastinating.

The pressureWhat it actually is
Copper retirement and forced fibre upgradesA real, scheduled programme affecting premises progressively. When it reaches your address, the analogue service on that copper ends. This is the hard deadline and it is site-specific.
ISDN and legacy product retirementCarriers have been retiring legacy business products on published timelines. These end dates are real but they are per-product and per-carrier, so check yours rather than a headline.
PBX end of supportSofter and often more urgent in practice. A system out of vendor support still works until it does not, and then there are no parts and no patches.
"You must move immediately"Usually a sales position rather than a fact. Verify your own site's date and your own product's end date before accepting a timeline from anyone, including us.
The one item worth acting on early

Not the phone system — the lift phone, the fire panel dialler, the EFTPOS backup line, the alarm dialler and the fax. Every business that gets caught by a copper cutover gets caught by one of those, not by the desk phones. They are on analogue services nobody has thought about for a decade, they are frequently compliance-relevant, and they need a specific replacement plan rather than an assumption that the phone project will pick them up. Make that list now, whatever else you decide.

Two Migration Paths, and the Seat Count That Decides

Once doing nothing stops being available, there are exactly two paths.

Path A: SIP trunk the existing PBXPath B: move to a platform
What you doKeep the box, replace its circuits with SIP trunksRetire the box, put the call control in the provider's platform
What it preservesThe call flow, the handsets, the muscle memory, the sunk costNothing physical — the call flow is rebuilt, usually better
What it leaves in placeThe single point of failure, the site visits, the end-of-support clock, the version
Where it genuinely winsA recent, supported, well-understood PBX with real remaining life; heavy customisation that works; a business that must not change anything this yearAlmost everything else, and every case where remote work, elasticity or AI matters
The honest riskYou spend money to keep an asset alive and face the same decision in three years with an older assetA migration project, retraining, and a period of everyone learning new habits

The seat count that decides. Below roughly fifty seats, keeping a PBX alive on SIP trunks is an expensive way to defer the cloud decision rather than a way to avoid it — the hardware, the support contract and the expertise all have to be funded by a small number of users, and the per-seat maths does not work. Above roughly one hundred seats, owning the platform can start to win again on pure cost, particularly where a business has heavy customisation, in-house expertise and stable requirements. Between those two figures it is a genuine judgement call and it turns on whether you have the expertise in-house, not on the technology. Anybody who gives you a single answer without asking your seat count is not doing the arithmetic.

Which Era Are You In? A Five-Minute Diagnostic

Answer these from observation, not from what you were told when you signed.

QuestionIf yes, you are probably in...
Is there a physical box in a cupboard that, if it died, would stop the phones?Era two or three. This one question resolves more cases than the other five together.
Does adding a new starter require somebody external, or a wait longer than an afternoon?Era two or three. Era four is a portal field.
Is there a fixed number of calls you can take at once, and does somebody know that number?Era one, two or three. In era four nobody knows the number because nobody provisions it.
Can a staff member take their extension home on a laptop or phone, with the same number?If no: era one, two or three. If yes, natively and without a VPN: era four.
Are call recordings, transcripts and summaries available in the platform without a second vendor?If no: era three at best. This is the clearest single indicator of era four.
When the site loses internet or power, do inbound calls still reach somebody?If no: era two or three. Era four fails over because the number was never at the site.
The awkward answer

A lot of businesses answer these and discover they are in era three while believing they were in era four, because a provider modernised the transport and described it as a cloud migration. That is the single most common misdiagnosis in this market. It is also the cheapest to correct, because the hard part — accepting that voice is packets and that upload capacity matters — is already done.

What Comes After Era Four

Predicting a fifth era would be speculation, so instead here is what is observably happening inside era four, which is more useful and less embarrassing to be wrong about.

🤖

The AI layer stopped being a feature

Transcription, summarisation, scoring and answering are moving from add-ons to defaults. The interesting question is no longer whether a platform has AI but whose infrastructure it runs on and what happens to the audio.

🔗

The phone system stopped being a phone system

Voice, SMS, chat, video and the customer record converge into one queue and one history. The word "phone" is becoming the least accurate part of the product name.

🧩

Open interfaces became the differentiator

When every platform has the same features, the difference is whether your own software can reach them. APIs moved from a technical detail to a procurement question.

🇦🇺

Where it runs started to matter again

Data sovereignty, latency and regulatory transparency have made "in whose data centre" a live question after a decade of nobody asking. This is the loop closing back to era one, where location was everything.

What to Do About It

Tell us which era you are in and we will tell you what we would actually do

Send us your seat count, whether there is a box in a cupboard, and what your carrier has told you about your site. If the honest answer is that you should sit tight for eighteen months, that is the answer you will get.

Get Started Or call 1300 881 662
The summary

Four eras: copper with one call per line and every feature metered; ISDN with a box in your building and capacity as a hard ceiling; VoIP, where the transport modernised and the box frequently stayed put and quality became your problem; and Voice over Cloud, where there is no box, capacity is elastic, any device is an extension and AI is native because the audio is already in the right place. VoIP is how voice travels. Voice over Cloud is where the system lives. They are two decisions and they get sold as one. Copper retirement removes the option of doing nothing, leaving two paths — and below about fifty seats, keeping the box alive is an expensive way to defer the decision rather than avoid it. Start with the lift phone and the fire panel, not the desk phones.

Related reading: what a cloud phone system is for era four in detail, hosted versus on-premises PBX for the path A versus path B decision with numbers, SIP trunking explained for path A specifically, VoIP versus landline for the era one comparison, and copper disconnection and forced fibre upgrades for the deadline that applies to your address.

Frequently Asked Questions

What is the difference between VoIP and Voice over Cloud?
VoIP is a transport technology and Voice over Cloud is a delivery model, which means they answer different questions and are separate decisions that the market habitually sells as one. VoIP describes how the voice travels: as data packets across an IP network rather than as an electrical signal down a copper pair or as a channel on a digital circuit. Voice over Cloud describes where the system itself lives: the call control software runs in the provider's data centre, and your handsets and apps connect to it over the internet, so there is no box on your premises. The practical consequence is that you can absolutely have VoIP without having cloud, and thousands of Australian businesses do exactly that — they replaced their ISDN circuits with SIP trunks somewhere around 2014 and left the PBX sitting in the same cupboard it had always occupied. That configuration is genuinely VoIP and genuinely not cloud. It still has a single physical point of failure in one building, on one power supply and one internet connection; changes still require somebody who understands that box; and the end-of-support clock on the hardware is still running. Businesses in that position are frequently told they are on the cloud, which matters because they then stop planning for the failure mode they actually have.
How do I tell which generation of phone system my business is running?
Six observable questions settle it, and you should answer them from what you can see rather than from what you were told at signing. One: is there a physical box somewhere on your premises that, if it died, would stop the phones? If yes you are in era two or three, and that single question resolves more cases than the rest combined. Two: does adding a new starter need somebody external, or take longer than an afternoon? If yes, era two or three — in a cloud platform it is a field in a portal. Three: is there a fixed number of simultaneous calls you can handle, and does somebody in the business know that number? If yes, you are in one of the first three eras, because in era four nobody knows it since nobody provisions it. Four: can a staff member take their extension home on a laptop or mobile, with the same number, natively and without a VPN? If no, eras one to three. Five: are call recordings, transcripts and summaries available in the platform itself without a second vendor? If no, era three at best — this is the clearest indicator of era four. Six: when the site loses power or internet, do inbound calls still reach a human? If no, era two or three. The common outcome is discovering you are in era three while believing you were in era four.
Is the PSTN being shut down in Australia and what does that mean for my business?
Australia's copper network is being progressively decommissioned and the PSTN and ISDN services that ride on it are being retired with it, which removes the option a great many businesses were quietly exercising — doing nothing indefinitely. Be precise about the pressure, though, because it is routinely overstated by people selling and understated by people procrastinating. The hard deadline is site-specific: copper retirement and forced fibre upgrades roll through premises progressively, and when the programme reaches your address the analogue service on that copper ends. Separately, carriers have been retiring legacy business products on published timelines, but those dates are per-product and per-carrier, so check the ones that apply to you rather than reacting to a headline. A third and often more urgent pressure is softer: a PBX out of vendor support keeps working until it does not, at which point there are no parts and no patches. The item genuinely worth acting on early is not the desk phones. It is the lift phone, the fire panel dialler, the alarm dialler, the EFTPOS backup line and the fax, because those sit on forgotten analogue services, are frequently compliance-relevant, and are what actually catches businesses out at a cutover. Make that list now regardless of what you decide about the phone system.
Should I put SIP trunks into my existing PBX or move to a cloud platform?
It depends on seat count and on whether you have the expertise in-house, and any provider who answers without asking your seat count is not doing the arithmetic. Below roughly fifty seats, keeping a PBX alive on SIP trunks is an expensive way to defer the cloud decision rather than a way to avoid it: the hardware, the support contract and the specialist knowledge all have to be funded across a small number of users, and the per-seat economics simply do not work. Above roughly one hundred seats, owning the platform can start to win again on pure cost, particularly for a business with heavy customisation that genuinely works, in-house expertise to maintain it, and stable requirements that are not about to change. Between fifty and one hundred it is a real judgement call that turns on capability rather than technology. SIP trunking the existing box makes sense where the PBX is recent, supported and well understood, where customisation would be expensive to reproduce, or where the business genuinely cannot absorb a change this year. What it does not do is remove the single point of failure, the site visits for changes, or the end-of-support clock — so the honest framing is that you are buying time at a price, and you will face the same decision in about three years with an older asset.
Why did call quality become a problem when businesses moved to VoIP?
Because responsibility moved along with the technology, and nobody announced it. On analogue and ISDN services, call quality was the carrier's responsibility and was effectively a solved problem — the network was purpose-built for voice and voice was all it carried. Once voice became packets travelling over a general-purpose internet connection, quality became a function of things the customer owns: your internet service, your router, your local network and above all your upload headroom. When calls break up, the carrier can accurately report that the service is within specification, because it is; the problem is in the space between the specification and what voice actually needs. This is the origin of every jitter, latency and one-way-audio conversation that has happened in an Australian office since roughly 2008. The practical implication is that businesses shop on the wrong number: the download figure is largely decorative for voice, while upload is what determines what happens when several people are on calls and somebody starts a video meeting. It is also why the recent nbn wholesale changes lifting business services from 40 to 100 upload matter more than the headline download increases, and why a provider who supplies both the network and the platform can diagnose these faults when two separate providers each correctly certify their own half.
What does Voice over Cloud give a business that a hosted PBX or VoIP setup does not?
Eight things, and the last is the one that has changed fastest. There is no box to own, so no hardware refresh cycle, no end-of-support cliff and no single cupboard whose flooding takes out the phones. Capacity is elastic rather than provisioned, so the eleventh simultaneous call is not a failure and you stop paying year-round for a peak that happens twice a year. Any device is an extension — desk phone, laptop, mobile or browser share one number and one call flow — so location stops being a property of the phone system. Billing is per user per month, so cost tracks headcount rather than infrastructure. Failover is automatic because the number never lived at your site. Provisioning is self-service, which collapses the cost of change and therefore changes what a business is willing to try. Upgrades are continuous rather than version migrations, so you do not run version 6.2 of anything. And AI is native rather than bolted on: transcription, summaries, call scoring and AI agents work because the audio is already in the platform. An on-premises system can only get those by exporting audio to a second vendor, which adds an integration and a data-handling question. The gap is not that cloud is newer — it is that the audio is in the right place.
Where did the term Voice over Cloud come from?
It was coined in 2017 by an Australian provider to name something that existing vocabulary was failing to describe. The problem was real: by the middle of the 2010s the word VoIP covered two completely different situations — a business running SIP trunks into a PBX it owned, and a business whose entire call control lived in somebody else's data centre — and the industry had no clean way to distinguish them. Hosted PBX was closer but carried the implication that a traditional PBX had simply been lifted into a rack somewhere else, which understated what had actually changed about elasticity, device independence and the pace of platform improvement. Voice over Cloud named the delivery model rather than the transport, and the term spread from one company's product language into wider Australian use over the following couple of years as other providers adopted it. That history is worth knowing for a practical reason rather than a sentimental one: the term exists precisely because the VoIP-versus-cloud confusion was costing businesses money, and it remains the fastest way to establish which of the two situations a supplier is actually describing. If a provider cannot tell you whether they are selling you a transport change or a delivery-model change, that is itself a useful answer.

What to Read Next

Your next reads

Uniden Voice Over Cloud logo

Australia’s smartest AI-powered cloud phone system — Australian owned, Australian hosted, Australian supported. unidenvoice.com | 1300 881 662