[ September 6, 2026 by Rob Kurver 0 Comments ]

The Capability Was Never the Problem

The capability was never the problem. Coordination is. Six years of working groups, roundtables, partnerships and research keep pointing at the same finding — and CASA26 in Amsterdam is built to test it.

CASA26 — The Capability Was Never the Problem. The market already knows what to build. What it cannot do alone is agree on what is true. Intelligence, Trust, Transformation. Amsterdam, 20–23 September 2026.
[ V A L U E   M O V E S   A L O N G   O N E   A R C ] 01 Connectivity Carrying it 02 Conversation Where CPaaS won 03 Intelligence What was meant 04 Insight What gets done [ T W O   D A Y S ,   T H R E E   T E S T S ] MONDAY 21 Intelligence Where AI already ships and gets paid for. TUESDAY 22 · AM Trust Who can be trusted to carry it. TUESDAY 22 · PM Transformation What it takes to move.

Two weeks ago I was on a stage in Nusa Dua at BATIC, Telkom Indonesia’s annual international conference, and the programme had opened with four framing questions. The fourth one was this: how do we turn next-generation infrastructure into inclusive AI-era communications services?

I hadn’t built my session to answer it. It is what CPaaSAA has been working on for six years — the working groups where practitioners solve the problems no single company can fix alone, the Talks and the closed-door roundtables, the ecosystem partnerships, the applied innovation work we now run with Sandbox Industries, and the research that comes out of all of it. The question acquired a number that week. It has had our attention rather longer.

The infrastructure half of it is being answered already. Capacity, subsea, data centres, sovereign cloud — going up across Asia at a pace Europe should find uncomfortable. And the ambition above the infrastructure is settled too. Every operator strategy in the world now contains the words orchestrator, intelligence, AI.

What is not settled is what actually gets sold, and whether anyone can buy it.

That gap is the reason CASA26 exists in the shape it does, and it is worth explaining before a hundred and fifty people get on planes to Amsterdam.

What six years keeps telling us

The clearest version came from the roadshow this year. Five cities, around a hundred conversations, more than fifty organisations engaged. We went out to map the market and match innovation to need.

We came back with something we were not looking for. The market already describes its own problem, unprompted, in almost the same words in every room: the capability exists, and it will not assemble itself. Operators asked us, repeatedly, how they plug into innovation happening outside their own walls quickly and without betting a flat budget on it. They hold proximity, regulatory trust and the customer relationship. What they lack is a low-risk route in.

Meanwhile the platforms are rewriting their own category in real time — dropping the CPaaS label, repositioning as orchestration layers, appointing new leadership aimed at an AI economy — and all of them are spending inward.

The pattern held everywhere. The capability is not the problem. Coordination is.

That is an uncomfortable finding, because you cannot buy your way out of it and you cannot announce your way out of it either. Going down the stack is straightforward: you already know exactly what you need, because you are your own most demanding customer. Going up, you are guessing at a buyer you have never been. That asymmetry explains most of what happened in this market in the first half of the year, and it explains why the CPaaS versions of suite products read thinner than the incumbents’ even when the engineering is better.

You cannot guess your way up the stack. Most of this market can no longer afford to buy its way up. What is left is committing your way up — together, with named customers, shared risk and published evidence.

Which is a thing you do in a room, not in a press release.

Two days, one argument

CASA26 is built as three tests of that finding. Not three themes. Three tests.

Monday morning is the macro view, and it is deliberately three people who do not share a vantage point. Ivan Ostojic of McKinsey ran commercial at a CPaaS at scale before advising the operators trying to capture the same value, so he has watched the same budget decision from both sides of the table — where AI spend is actually landing, and which programmes survive their first renewal. Leonard Lee of neXt Curve then reads the same market from outside it: where the architecture question is being settled, who ends up owning the layer above the model, and why most of those decisions are being taken outside Europe. And Neelam Sandhu, who built her CASA25 keynote around intelligent engagement before the term had travelled anywhere, comes back to say what has actually changed since inside companies that already own a communications stack they have paid for.

Kunal Shukla of Vonage follows with the platform version of the same problem — network APIs and AI converging faster than either roadmap assumed, and what a large CPaaS does when it has to sell both to enterprise customers who want neither in the abstract.

Intelligence runs through Monday afternoon. The question is not whether AI works — that argument is over. It is what it is worth. Radisys will talk about AI voice running inside the operator’s own core, with named deployments and real numbers rather than a company overview. ElevenLabs and KPN will demonstrate voice intelligence live rather than describe it.

Then Žiga Lešnik of Simon-Kucher chairs the session I think is the most useful hour on the programme, with Meetric, Radisys and Sinch’s chief product officer: what do you actually charge for intelligence? Everyone agrees value is moving up the stack. Almost nobody has changed their price list. This industry priced itself per seat and per message for two decades, and it is now starting to price against the cost of a token — which reprices the product, the margin and, in time, the company.

Sam Wilson of 8×8 then argues the deliberately contrary case, that nobody cares about the AI at all, only about what you do with it and how you sell it. Daan de Wever of Dstny makes the opposite kind of argument — sovereignty as a product category rather than a compliance cost, built in Europe for buyers who cannot put regulated conversations through a centralised model, and priced accordingly. Dave Michels closes the day with both of them, plus Vonage and Sinch, asking what UCaaS, CCaaS and CPaaS still mean if anything, and what happens to the companies named after them.

Trust runs through Tuesday morning, and it exists because none of Monday is sellable if you cannot prove who is on the other end of the conversation.

Aoife Condren, CMO of Proximus Global, opens it — network APIs as a communication fabric rather than a product line, from the operator that has committed hardest to it. Andrew Collinson then launches our mobile network API research alongside STL Partners — eighty executive interviews, twenty-one markets assessed for their legal position — with the evidence, the commercial models, and the parts that are not working. Then Paolo Fodor of Vydens, who has built a retail product natively on Open Gateway, on what network intelligence does in physical stores — and, unusually, addressing consent on stage rather than leaving it to the legal annexe. Amy Cameron chairs the GSMA panel that follows, under a title I would have chosen myself: enterprises don’t buy APIs, so what is the industry actually selling?

The GSMA built that hour, and built it the right way round. They could have sent someone to talk about standards progress. Instead they brought companies that have shipped on Open Gateway and have customers — Vydens in retail, WeVerify in identity verification — and put them on a stage with Vodafone and Proximus Global. That is the difference between a standards update and evidence. They then take the whole of Wednesday for a closed, invitation-only roundtable of operator champions, working through Open Gateway implementation in regional clusters. Operators only, no vendors, and not a session we run.

The morning’s second half narrows to identity and provenance. XConnect and Honey Badger will show network signals working in production against fraud, sold to lenders. Then a panel on the question this industry keeps postponing: who pays for the conversation record, and under whose law does it sit? Crexendo, Tresic and Simwood, with no agreement between them about where the money sits — and a carrier argument that platform players cannot retrofit, because provenance is decided at the point of origination or not at all.

Transformation is Tuesday afternoon, and it is the “so what do we do” block. Shilpa Strong on how anything actually moves inside a large organisation, and where proposals go to die. Chris Zock of Sandbox Industries on what five cities of conversations produced — capital wants into this sector and cannot find the front door. Dean Bubley chairs an argument about sovereignty and who should hold the data, with Deutsche Telekom’s wholesale arm, Telin, Alianza and Google Cloud. And Andrea Reynolds on the same coordination problem from the founder’s side of the table.

John Banta of Blue Venture Fund makes the case earlier, on Monday morning, and it should interest every operator in the room: competitors who co-invest together, why it started in healthcare, and what it has actually returned.

And then, near the close, real early-stage founders pitching to real investors rather than a demo judged by analysts. The point is not the format. It is that the coordination problem tends to get solved by companies this size, and the room should leave knowing their names.

The part I am most interested in

On Monday before lunch, the room does something it does not usually do at an industry event. It commits.

Andrew Collinson will set out two axes — how telco propositions fare against non-telco, and how AI reshapes the shape of the industry — with short positions from Nick Lane and Raúl Castañón of 451 Research, and Amy Cameron in the discussion. Then we vote. Not a survey. A room of senior people putting their hands up for a position in front of their peers.

On Tuesday evening, last session of the event, we read the result back and ask what would have to be true for each outcome.

That is the whole design, honestly. Everything between those two moments is evidence.

Around the sessions

One thing I am particularly pleased about. Every panel across the two days is chaired by someone independent — Andrew Collinson of Connective Insight, Amy Cameron and Miriam Sabapathy of STL Partners, Dave Michels of TalkingPointz, Dean Bubley of Disruptive Analysis, Raúl Castañón of 451 Research, Nick Lane of Mobilesquared, and Žiga Lešnik of Simon-Kucher on the pricing session. No vendor chairs a panel they are sitting on. That is harder to arrange than it sounds, and it is most of the difference between a session that goes somewhere and one where four companies take turns describing themselves. CX Foundation and TelecomTV are both covering the two days.

The expo runs throughout and lunch is served in it, so you will end up in front of the sponsors whether you plan to or not. Both mornings run without a formal coffee break — there was more worth putting on stage than there was room for — so the days are full. Coffee from nine, and the afternoons breathe.

The social programme starts before the programme does. Sunday evening there is a canal boat, hosted by Sandbox Industries: investors, founders and operators, no pitches and no running order. You add it to your ticket separately, because the boat is not large. Monday night is the party at TufTuf Club, off-site, where most of the useful introductions actually happen. Tuesday we finish at the venue with drinks and proper food until seven, and after that the evening is yours.

What we are not doing

We are not going to settle this in two days. The consent mechanism that European operators told us does not exist anywhere is being built in India, Indonesia and Singapore right now, while reform of the European instrument runs past 2028. Assured connectivity does not go commercial in most markets until 2027. The buyer is quietly solving the problem without us, with passkeys and wallets, and is rarely in the room at events like ours — including ours.

None of that gets fixed by a hundred and fifty people in Amsterdam.

What can happen in two days is narrower and more useful: the people who would have to move together find out whether they agree on what is true. That is the prerequisite for everything else, and it is the one thing that genuinely cannot be done remotely, asynchronously, or alone.

The evidence says this market works. It is at the top of wait-and-see. What it needs now is nerve, and a few players willing to move at the same time.

See you on the 20th.


CASA26 runs 20–23 September at the Tobacco Theater in Amsterdam. Closed-door, members and partners. The full programme — every session, every speaker, every time — is at casa26.amsterdam/casa26-schedule, and it is kept current as things move.