— Eric S. Raymond, The Cathedral and the Bazaar, 1999
Eric Raymond’s quote is the case for open source in seven words. A closed shop has a few dozen engineers looking at the code. An open shop has thousands, including people who use it in ways the original authors never imagined. Bugs that might take a proprietary vendor months to find and fix get spotted and patched in days, because somewhere in that crowd is someone who spots what everyone else missed. The eyeballs only count when the project is alive, which is why active maintenance and curation matter as much as openness itself.
Today, 27 years since publication, that prediction has held in nearly every infrastructure layer that matters. Linux runs the internet. WordPress runs nearly half the web. Telephony is no different: Asterisk, FreeSWITCH, and Kamailio aren’t scrappy alternatives anymore. They’re the dominant code in the SMB and MSP segment, deployed in the millions, hardened by every weird and wonderful configuration the world has thrown at them. A proprietary vendor’s QA team tests what they thought to test. Open source has been tested by reality.
Commercial sense favours open source and the technology is mature enough to take it seriously. The proprietary vendors keep doing the marketing for the open source side with every price rise and forced migration. The MSP channel hasn’t pivoted. Why?
Because the economics favour open source for the MSP, but the sales motion favours proprietary for the salesperson. Those aren’t the same thing.
I have worked in the telecoms industry for 18 years. In that time I have watched the cloud-comms cartel raise prices, sunset products, restructure channel programmes, and gradually absorb the margin that used to belong to MSPs. I have watched them buy and sell each other, tearing apart the fabric a customer once bought into. I have watched the open source side get better, faster, more stable, and more capable every single year.
A bit of context on where this view comes from. 20tele.com has been delivering managed hosted PBX through THEVOIPEXP for years. Fully maintained, fully supported, configured and tuned by us for customers who want the phone system to be someone else’s problem. That’s one shape of the market and it works. FreePBX UK is the other shape. Self-service, purpose-built hosting, configured for the use case, ready for the customer or their reseller to take from there. Same underlying technology, two different doors into it. The managed door for hands-off, the self-service door for hands-on. Widening the net is the whole point.
A typical cloud-comms reseller collects recurring revenue with a clean quote-to-cash flow. The vendor does the heavy lifting. The MSP does the talking. FreePBX requires you to actually understand the product, host it, support it, and sell on value rather than by brochure. Most MSPs are sales-led, not engineering-led. They’ll take the easier margin every time, even when it’s a smaller margin, because the alternative requires changing what the business is.
An MSP that resells cloud-comms is in the relationship business. The product is interchangeable, the margin is set by the vendor, and the value the MSP adds is account management. An MSP that runs open source PBX is in the engineering business. The product is theirs to shape, the margin reflects the work they put in, and the value they add is the curation itself. These are two different companies wearing the same logo. Pivoting from one to the other isn’t a strategic decision. It’s a rebuild.
There is a generational filter as well. The MSPs who came up through traditional telecoms, ISDN, SIP trunks as a product, on-prem PBX, understand the stack and could pivot tomorrow if they decided to. The MSPs who came up through cloud-comms in the reseller channel don’t have the skills in-house. They aren’t choosing proprietary, they’re trapped by it. Retraining a sales-led shop into an engineering-led shop is a multi-year cultural project, and most owners aren’t going to start that at this stage of their career.
This isn’t a criticism of those owners. Reselling is a perfectly legitimate way to make a living, and the cash flow is predictable in a way engineering-led businesses rarely are. But it does mean the channel as a whole is structurally biased against the pivot. The MSPs who could move easily often have no commercial reason to. The MSPs who have every commercial reason to often can’t move easily. That asymmetry is why the shift will be slow, and why the early movers have a long runway before the rest catch up.
The strongest case for proprietary, and the one I have heard from MSPs more times than I can count, goes like this: single vendor, single bill, single support number. Sales people benefit from incentives and prizes, warm lead handouts, marketing development funds, training and accreditation. Boardrooms benefit from predictable margins, predictable renewals, little-to-no engineering risk. If something breaks, you escalate to the vendor and they own it. With open source, you own it. That’s a real argument and it should be respected, encouraged even in some cases. This isn't for everyone, and pretending otherwise helps no one. For an MSP whose customers are non-technical SMEs and whose differentiation is hand-holding rather than capability, sticking with proprietary is a reasonable choice.
The well-rehearsed "one-throat-to-choke" pitch is a genuine value proposition, not marketing fiction.
I have sat in those rooms and made that pitch myself, and I was good at it. I worked for BT Wholesale, then Gamma, and I went on to run a cloud-comms reseller business of my own. We sold a lot more than we do now, but marketing in those times was much easier. What did we actually own, though? Nothing. The end user and the vendor were in a love triangle, and we were the third in that relationship.
The case has weakened in the last few years and it is weakening faster now. The one-throat-to-choke pitch translates into single-point-of-failure, in commission-laden packaging. When a major vendor raises prices in a renewal cycle, when features get removed from a product you already sold, when an entire reseller programme gets restructured and your discount tier evaporates, the vendor relationship that was supposed to protect you becomes the thing eating you. Lock-in cuts both ways. The predictable margin gets less predictable every year, because the vendor has every reason to keep more of it for themselves as the channel matures.
Then there’s the reputation problem on the open source side. The MSPs who do go open source often do it badly. Undifferentiated FreePBX hosting, a race to the bottom on price with questionable support standards, SLAs not worthy of an acronym. They give the category a reputation for being cheap and cheerful rather than serious infrastructure. That’s the gap FreePBX UK is filling, but it’s also a large part of why the broader MSP world hasn’t rushed in. They look down on what they perceive to be the low end, decide the category isn’t for them, and stay where they are.
The cheap and cheerful, low end isn’t what open source is. That’s what bad implementations of open source look like. Done properly, open source PBX matches or beats proprietary on reliability, flexibility, and cost. Done improperly, it’s a cheap droplet with no monitoring, no patching schedule, and a support inbox nobody reads. Same technology in both cases. Curation is where MSPs ought to be making a difference and earning their keep, which is exactly the part too many of them have outsourced to a cloud-comms vendor.
Now, let's look at the numbers. A typical hosted seat from a major UK reseller sits somewhere between £8 and £15 per user per month, depending on feature tier and call bundle. Your mileage may vary, but let me make the point. The hosting costs are buried somewhere in that number. A FreePBX hosting tier on FreePBX UK starts at £15 a month for the entire system, not per user. At 20 users, that’s 75p per seat. At 50 users, it’s 30p. The cost gap isn’t a margin difference, it’s an order of magnitude. The MSP that figures out how to deliver that cost structure with proper support on top has a pricing weapon that competitors with reseller agreements can't even pretend to match. They have to fund the vendor’s overhead. You don’t.
You may be thinking, where’s FreePBX UK's reseller margin? Where’s the channel programme? Where’s the discount tier?
There isn’t one, there isn’t going to be one, and I will be clear about why.
The price you see is what the product is worth, given what it costs to provision, host and support the infrastructure properly. There is no fat baked in waiting to be carved out for a middle layer. Building a reseller margin into the price would mean charging the customer more than the product is worth, so a third party can take a cut for doing nothing the customer asked for. That isn’t a business model, it’s a tax. And it’s exactly the tax the proprietary world has been charging for 20 years.
20tele has the UK Sangoma partnership, the rights and a licence for hosting FreePBX. We’ve earned them and we use them. We aren’t giving the margin away for the privilege of someone else putting their badge on what we’ve built.
What I will do is talk volume. Bring real numbers, bring them consistently, bring them with the engineering discipline to support what you sell, and there’s a conversation to be had. But it’s about volume, not margin. My job is to grow the base and put pressure on suppliers to bring costs down. Fattening the pie for resellers is the opposite of what I’m trying to do.
This is a deliberate cultural shift, and I know it. The UK telecoms channel runs on resellers expecting 25% to 50% margin baked into wholesale pricing as a matter of course. That’s how the industry has worked for decades. FreePBX UK isn’t playing that game. The MSP makes their money on the things the MSP actually does: deployment, configuration, integration, training, SIP trunks, hardware, ongoing administration, support, and maintenance. Everything except the software and the hosting. The reseller margin on those two things is zero, and the customer keeps the difference.
There’s also a stickiness argument that doesn’t get made often enough. A customer on a proprietary platform is one renewal email away from leaving. A customer on a properly run open source PBX has their dial plan, their queues, their IVR, their phones, their integrations, and their habits all built around a system they understand and control. Switching costs are real, and they accumulate. Open source customers don’t churn for cheaper alternatives because they’re already on the cheaper alternative. They churn for worse service, which is something the MSP controls.
The UK angle matters too, and it’s worth saying out loud. Not Rule Britannia. It’s a question of where the service layer actually lives. Customer data hosted in the UK, by a UK company, on UK infrastructure, with support staff in UK working hours. The code can come from anywhere, and FreePBX itself comes from a partner we work closely with in North America. The people, the hosting, the support, the commercial relationship, those belong here. With the BT PSTN switch-off forcing every business in the country to make a decision about their phone system anyway, the question of where the service actually lives has become a procurement question. UK businesses want UK answers to UK questions, even when the engine underneath was built somewhere else.
So will the channel pivot? Yes. Two forces will push it and both are accelerating.
The proprietary vendors will keep doing the work for me. Every price rise, every forced migration, every “we are sunsetting this product” email sends another batch of buyers looking for alternatives. The buyer goes first. The MSP follows the buyer. Customers don’t write think-pieces about software philosophy. They get a renewal quote, choke on it, and ask their MSP what else is out there. If the MSP has nothing else to offer, the customer finds an MSP who does. Pressure builds from underneath, customer by customer, quote by quote, and at some point it tips.
The skills are getting easier to acquire. The Heqet ISO and IN1CLICK are part of that, making FreePBX easier than ever to install. So is the broader trend of FreePBX getting easier to maintain with every module update. The learning curve is being flattened by AI tooling, and recent innovations like Frogman demonstrate what the future of phone system administration will look like. The barrier to entry for an MSP to add open source PBX to their stack is dropping every year, which means more of them will add it as a second option alongside their cloud-comms reseller deal, then start noticing the margin is better, then start noticing the customers are stickier, then stop pushing the reseller deal.
That’s how channel pivots actually happen. Not by announcement. By drift. The reseller deal doesn’t get cancelled, it gets neglected. The open source line doesn’t get launched, it gets gradually expanded.
Most MSPs won’t lead this shift. They’ll follow it once it’s safe, which is the same thing MSPs always do. The competition for FreePBX UK isn’t them. It’s the proprietary vendors’ marketing budgets and the inertia of “no one got fired for buying the obvious thing.” That inertia is real and it’s well-funded, but it has a shelf life. Every few months the proprietary case gets harder to make, because every few months the open source stack gets better and the proprietary price gets worse.
What’s actually at stake is whether the open source PBX ecosystem can produce enough credible, well-curated, MSP-grade offerings before buyers give up and resign themselves to whichever proprietary cloud is least bad. That’s the actual race. Not open source versus proprietary in the abstract, but curated open source versus the path of least resistance.
There’s one more force pulling all of this in the same direction, and it’s going to accelerate everything I’ve written above. The lazy prediction was that AI would entrench the proprietary giants, because only they could afford to train the models.
Wrong. Open weight models are within touching distance of frontier proprietary ones, the ecosystem of fine-tuners and integrators is exploding, and the most interesting work in the field is happening in public repositories rather than behind corporate firewalls. The world is asking the same question of every vendor: show me your workings so I can trust what you have or build it better. The eyeballs argument is winning the AI layer too, on a faster clock than it won the operating system layer or the database layer.
And before anyone raises it: yes, AI can scan open source code for vulnerabilities. So can attackers. The difference is that AI scanning open code gives defenders the same chance to find the bug first, because they have the same access. AI scanning closed code only happens if the vendor decides to pay for it. Security through obscurity was already discredited. AI just buried it.
What this means for Software-as-a-Service is bigger than most vendors are willing to admit. AI changes who can read code, and who can extend it. A small MSP with one good engineer and a competent AI assistant can now do work that used to need a team of five. The skills barrier that protected proprietary vendors for decades, the assumption that customers would never have the in-house expertise to run software themselves, is dissolving as we speak. The proprietary SaaS pitch has always been “we run it so you don’t have to.” That pitch gets weaker every month AI makes running it cheaper.
Linus Torvalds said it best decades ago:
Open source has been showing the code for as long as it has existed. Proprietary has been showing the invoice.
Open source isn’t the poor relation to SaaS. It’s the manual. The DNA. The beating heart of the SaaS itself. Most major SaaS products on the market are built on open source code at the foundation, then wrapped in a proprietary control panel and sold back to the customer at a markup. The customer isn’t paying for the software, they’re paying for the wrapper. AI is making wrappers cheaper to build and harder to defend. What survives is the underlying code, which is open, and the curation layer on top of it, which is where the real value has always been. The proprietary middle is being squeezed from both ends.
That’s why the bet on open source PBX isn’t contrarian. It’s a bet on the direction every infrastructure layer has already moved in, accelerated by an AI cycle that strengthens open source rather than weakening it. The MSPs who understand this are positioning themselves on the right side of a trend that has already played out in every adjacent market. The MSPs who don’t are betting their next decade on reseller agreements with vendors whose grip is loosening. That bet may pay out for a few more years. Eventually it won’t.
It’s the bet FreePBX UK and 20tele are making. Sam Knowles and I have bet our careers on it. The eyeballs have done their work. The code is hardened, the feature set is mature, the deployment story is solved. What was missing was a UK-shaped front door, hosted in the UK, priced for the UK market, supported in UK hours, with the curation done up front so the customer or their reseller doesn’t have to do it themselves. That’s what we’ve built. Trust that’s earned, by 18 years in telecoms, nine of those as a Sangoma partner.
I am set on this. You might disagree, and if you do, I’d genuinely like to hear why. But I’ve sat with this a long time and I’m betting on it with my own business, not someone else’s. We’ve thought about it, built for it, priced for it, and committed to it.
Bets Are Open.
Kieran
This blog post was published as a LinkedIn article here.
Bets Are Open