What Actually Changed
Let us start with the facts and dates, because a lot of the commentary around this has been vibes rather than specifics, and you cannot have a useful conversation with a customer using vibes.
| Change | Detail | Timing |
|---|---|---|
| Extension fair use policy | A published maximum number of extensions per simultaneous call licence size, formalising an upper bound where “unlimited extensions” had been widely understood | Ratios published with effect from late October 2025 |
| Informational warnings | Systems over the limit surface a notice in the console | From V20 Update 8 |
| Renewal warnings | Systems over the limit get a warning at renewal in the ordering system, indicating an upgrade or extension removal will be required during 2026 | From 1 January 2026 |
| Enforcement | Enforcement begins, with a grace period; systems still in violation risk losing access to technical support | From 1 April 2026 |
| Edition consolidation | Enterprise Plus retired; Enterprise renamed AI Edition; lineup becomes Free, Basic, PRO and AI | 23 April 2026 |
| Trial keys | Issued as 8SC PRO rather than 4SC PRO, with validity reduced from two months to one | From late April 2025 |
| Hosting | 4SC keys can no longer be hosted by 3CX; existing hosted 4SC keys renew as normal, but if one lapses it must be upgraded to 8SC to recover hosting | Current |
To be fair to 3CX
Every one of these is a defensible commercial decision. “Unlimited extensions” against a concurrency-priced licence was always an arbitrage that some deployments exploited hard, and formalising a ratio is not unreasonable. The edition consolidation genuinely simplifies a lineup that had become confusing. This page is not an argument that 3CX behaved badly. It is an argument about what happens to your business when the terms of the product you resell can move underneath you, and about whether the model you are in is the one you would choose today.
The Extension Cap, With Arithmetic
The policy ties the maximum number of extensions to the simultaneous call licence. The published ratios land in the range of roughly five to eight extensions per concurrent call:
| Simultaneous calls | Maximum extensions | Effective ratio |
|---|---|---|
| 8 | 40 | 1 : 5 |
| 16 | 80 | 1 : 5 |
| 32 | 176 | 1 : 5.5 |
| 64 | 384 | 1 : 6 |
| 96 | 576 | 1 : 6 |
| 128 | 768 | 1 : 6 |
Now put a real Australian deployment against it. Take a school you installed four years ago: 240 extensions — every classroom, the staffroom, the sick bay, the groundsman's shed — running comfortably on a 32SC licence, because a school makes very few simultaneous outside calls. It has worked perfectly for four years and the customer is delighted.
That system is now 64 extensions over the cap
32SC permits 176. The site has 240. The options are: uplift to 64SC, which nearly doubles the licence cost for zero additional capability the school will ever use, or remove 64 extensions, which means walking into a school and taking phones off walls. Neither is a conversation you want to have, and both of them are your conversation to have — not the vendor's.
3CX has indicated the change affects roughly 15% of customers. That number is accurate and, for a partner, misleading — because the 15% is not randomly distributed. It concentrates precisely in the deployments where 3CX was the obvious technical choice and where partners built their local reputation.
Who It Hits Hardest
The profile is simple: many endpoints, low concurrency. If that describes your best reference sites, this is your problem more than it is the average partner's.
Schools and campuses
A phone in every classroom and office, a handful of outside lines in use at any moment. The textbook case for concurrency licensing, and the textbook case for being caught by a ratio.
Hotels, motels and clubs
A handset in every room. Concurrency that peaks at a fraction of the endpoint count. Ratios of 20:1 or worse are entirely normal in hospitality and are now well outside policy.
Aged care, clinics and health
Phones in wings, rooms, nurse stations and consult rooms. Low concurrency by design. Also the customers least able to absorb an unplanned licence uplift mid-year.
Retail and multi-site
Twelve stores, four handsets each, two calls at a time across the whole group. The extension count is driven by geography, not by call volume — and the ratio does not care.
Warehousing and manufacturing
Handsets on the floor for safety and coordination rather than for external calls. Frequently the highest extension-to-concurrency ratio of any vertical.
Strata, body corporate, community
Common-area and facility phones that exist to be available, not to be busy. Small budgets, annual committee approvals, no appetite for a surprise.
Notice the pattern. These are not marginal customers — they are exactly the sites that made the technical argument for 3CX compelling in the first place. The characteristic that made the product a good fit is the characteristic the new policy prices.
The Edition Consolidation
On 23 April 2026, 3CX retired Enterprise Plus, renamed Enterprise to AI Edition, and settled the commercial lineup at Free, Basic, PRO and AI. AI Edition carries the PRO feature set plus the AI capabilities — receptionist, personal assistant, agents, transcription.
Two things follow for a partner, and the second is more important than the first.
| Consequence | Why it matters to you |
|---|---|
| AI now sits in the top tier | Every AI conversation with a customer becomes an upgrade conversation. When a competitor includes AI answering in a mid-tier plan, your quote loses on the line item your customer is most curious about in 2026 |
| Editions can be renamed and retired | You built proposals, comparison sheets, internal pricing tools and customer expectations around a lineup that no longer exists. That work is not billable and you will do it again |
The AI positioning matters more than partners generally credit. Demand for AI call answering has moved fast in the Australian small business market, and it is now one of the first questions a prospect asks. If your answer is “yes, on the top edition, with a third-party provider connected”, you are explaining an architecture at the exact moment your competitor is describing an outcome.
The Structural Asymmetry
Here is the part that has nothing to do with 2026 and everything to do with why 2026 stung.
| In a licence-resale model | You own | The vendor owns |
|---|---|---|
| The customer relationship | ✓ | |
| First line support | ✓ | |
| The install and the config | ✓ | |
| The blame when it breaks | ✓ | |
| Licensing terms | ✓ | |
| Pricing | ✓ | |
| The roadmap | ✓ | |
| Whether the edition you quoted still exists | ✓ |
Everything in the left column is a cost. Everything in the right column is control. When those two columns sit in different companies, every change made in the right column is absorbed as work and as reputational risk in the left one — and you cannot price for a change you did not know was coming.
The trunk problem, which is the same problem
3CX is the PBX. The SIP trunks come from somewhere else. When a customer reports one-way audio or a failed transfer, you are the one who has to determine whether it is the PBX, the trunk provider, the customer's internet or the handset — and you do that with commercial leverage over none of them. Every hour of that diagnosis is unbilled, and the customer experiences the wait as your service, not as a boundary between three vendors.
Partner community discussion through 2025 and 2026 has centred on a related theme: price changes mid-quote, uplifts at renewal that need justifying to end customers, and partners who built their businesses on small deployments finding those deployments no longer support their partner status or economics. Whatever the merits in each case, the common thread is that the terms moved and the partner absorbed it.
Annual Licence vs Monthly Recurring
This is the comparison that actually decides things, and it is worth doing with your own numbers rather than accepting anyone's illustration. The figures below are illustrative and deliberately conservative — plug in your own.
| A 25-seat customer | Licence resale model | Wholesale service model |
|---|---|---|
| Revenue shape | One licence renewal a year, plus project work | Recurring monthly, every month, per seat |
| Your margin sits in | Licence markup and your own labour | The spread between wholesale and retail, ongoing |
| Support escalation | You, then a ticket queue offshore, then wait | You, then an Australian engineer who owns the platform |
| Trunk faults | A third party. Possibly a fourth | Same company as the platform. One escalation |
| Price certainty | Set by the vendor, changeable | Wholesale rate agreed with you |
| Business valuation | Project revenue is valued at a low multiple | Contracted MRR is the asset buyers actually pay for |
| What happens if you stop selling | Renewals continue for a while, then decay | The base keeps paying while it is supported |
1×
Licence revenue per year
12×
Recurring revenue per year
1
Escalation path, not three
0
Licence terms that can move
The last row of the table is the one that changes how people think. If you ever intend to sell your business, or to step back from it, the valuation conversation is almost entirely about contracted recurring revenue. A licence resale book does not read the same way on a balance sheet as a supported, contracted base — and the difference is usually larger than a year of margin. Our piece on recurring revenue for MSPs and ISPs works that through properly.
What a Real Partner Program Provides
“Partner program” is one of the emptier phrases in this industry. Here is the checklist we would use to judge one, including ours — ask every one of these of anyone courting you.
| What to demand | Why it matters |
|---|---|
| Wholesale pricing, not a referral fee | A referral fee is a one-off. Wholesale means you own the margin and the pricing decision, month after month |
| You keep the customer relationship | Get it in writing. A supplier that can go direct to your base is a competitor with your customer list |
| Second line support by engineers who own the platform | The difference between an escalation that ends and a ticket that circulates |
| Network and platform from the same company | Removes the trunk-versus-PBX diagnosis entirely. This is the largest single reduction in your unbilled hours |
| Australian support hours that match your customers' | A 7am fault in Perth is not an overnight ticket |
| Provisioning you can run yourself | Adding a user should not require a supplier ticket. If it does, your margin gets eaten by admin |
| AI, SMS, video and recording included, not tiered | Every capability behind a paywall is a sales conversation you have to win twice |
| Hardware supplied and provisioned | Zero-touch handsets mean a site install measured in hours, not days |
| Pricing you can hold for a customer term | You cannot sell a three-year contract on wholesale rates that can move next quarter |
The one that catches people out
Ask specifically: “can you sell direct to a customer I introduced?” Ask for the answer in the agreement, not in an email. It is the single most consequential clause in any wholesale relationship and it is the one most often left comfortably vague.
Moving a Base Without Moving It All
The objection we hear most is reasonable: I have forty 3CX sites, I cannot move forty sites. Correct — and nobody should try. Every partner who has done this well has done it in the same order.
| Stage | What moves | Why this order |
|---|---|---|
| 1. New business only | Every new deal from today | Zero risk to your existing base. You learn the platform on customers with no prior expectations, and the recurring revenue starts immediately |
| 2. The forced conversations | Sites facing an extension cap uplift or an awkward renewal | These customers are already having a change conversation. You are choosing the destination rather than defending the increase |
| 3. The support-heavy sites | Whichever three sites generate the most unbilled hours | Biggest immediate margin improvement. You know exactly which three without looking them up |
| 4. The AI-curious | Customers asking about AI answering, SMS or better reporting | You are moving them toward something they want rather than away from something they have |
| 5. Everyone else, at renewal | The remainder, over 12–24 months | Renewal is the natural moment. No forced migrations, no drama, no weekend cutovers you did not budget for |
Note that stages one and two require you to move nothing at all. That is the point. You can test the model with real customers and real margin before making a single decision about your existing base — and if it does not suit you, you have lost nothing. The mechanics of an individual site move are covered in our 3CX to Uniden Voice partner migration guide, which is the how; this page is the whether.
How to Decide
Six questions. Answer them with your own numbers this week rather than at your next renewal, when you will be answering them under time pressure.
| Question | What the answer tells you |
|---|---|
| How many of your sites are over the extension ratio? | Count them today. Each one is a conversation you will have to initiate, and initiating it with a proposal is far better than responding to a warning |
| How many unbilled support hours did you write off last quarter? | Most partners have never totalled this. It is usually the largest single number in the decision |
| What proportion of your revenue is recurring and contracted? | If it is under half, you are running a project business that happens to sell phone systems |
| How many vendors sit between you and a resolved fault? | Every one adds a handover, and handovers are where hours disappear |
| Can you answer an AI question without saying “upgrade”? | In 2026 this comes up in most new conversations. It should not cost you the deal |
| If you sold the business tomorrow, what is the book worth? | Contracted MRR is what gets valued. Licence resale largely does not |
A fair closing note
3CX remains a capable product and plenty of partners will run profitable businesses on it for years. If your deployments are concurrency-heavy rather than extension-heavy, if your support load is low, and if your renewals are going smoothly, there is no crisis here and you should ignore anyone telling you there is. The argument on this page is for partners whose best sites are exactly the ones the ratio penalises, and who are carrying support for terms they cannot influence. If that is you, the thing worth changing is not the licence — it is the model.
If you want the wider commercial context first, the dealer program in full covers rates, support and onboarding, and the contact centre software buyer's guide covers the capability set your customers will be comparing you against.