A personal reflection on a quarter spent listening — five cities, a hundred conversations, and a question that changed along the way.

By the time I sat down to the last dinner of the quarter last week — a long table at Marv & Ben in Copenhagen — AWS and Aduna, Vodafone and Proximus, Vonage and Ribbon, Tech Mahindra, GSMA and neXt Curve around it, sponsored by Sandbox — I’d stopped counting the conversations. Five cities. Something close to a hundred of them. Three dinners built as working sessions rather than networking. More than fifty organisations engaged out of the hundred and fifty we’d mapped, and a stack of published pieces written on trains and in airport lounges along the way.

That last one was the best dinner I’d had in years — and I wasn’t the only one at the table who said so. But the food, as good as it was, wasn’t what made the night. It was the room: the selection of people was good enough that the conversation ran itself, and nobody needed to chair it. You could feel something click into place.

I didn’t set out to write a quarterly report. But somewhere between Paris in April and Copenhagen in June, the question I was asking changed — and that’s the part worth writing down.

The road was the method

The route was deliberate. Paris for the developers and a sovereign-AI conversation with a European operator. Madrid for the Aduna Summit and the real state of network APIs. New York for a dinner and the BT Exchange. London for DSP Leaders and the “move faster together” argument. Copenhagen, at TM Forum’s DTW, as the capstone. In between, more dinners and a steady drumbeat of research — market mapping, the AI-voice market, the network-API economics nobody enjoys doing.

Going on the road is not the efficient way to learn an industry. It is the only honest one. You can’t read where a market actually is from a slide deck or an analyst note; you read it from the third part of a dinner, when a CTO stops giving you the official line and tells you what’s actually keeping him up. So I went, and mostly I listened.

What I went to find out

I went out with a thesis and a genuine willingness to be wrong about it. CPaaSAA, together with our partners at Sandbox, has spent this year on one stubborn question: how do you connect the innovation happening at the edges of this industry with the capital and the operators who need it — and do it so it creates real value for everyone at the table, not just motion? The roadshow was the test. Engage every kind of player, in person, and find out where they really are.

They sorted into a few clear groups. And each one is in a very different place than the headlines suggest.

Where everyone actually is

The telcos and network operators are the ones holding both the real assets and the real anxiety. Revenue-rich and, for the most part, structurally slow — a “stay relevant in the AI era on a flat budget” problem that no amount of strategy-deck enthusiasm solves. What they have that hyperscalers can’t easily copy is proximity, regulatory trust, and the customer relationship. What they lack is a fast, low-risk way to plug into the innovation happening outside their own walls. That gap is the whole story.

There is also, for the first time in a while, a concrete reason for optimism. AI voice and agents were the hottest topic in every room I sat in this quarter — and they hand telcos something genuinely new. Attach an agent to a plain old phone call, and the most commoditised product in telecoms becomes a surface for intelligence. The infrastructure players are already building that bridge — the likes of Radisys and Mavenir are wiring AI straight into the call. The asset operators had quietly written off turns out to be a distribution channel for exactly the thing everyone now wants.

And plenty of operators are well ahead of that caricature. Across the quarter I had real conversations with a dozen of them — building programmable networks, leaning hard into network APIs, standing up AI-ready infrastructure, making serious moves in cloud communications, several with bold plays in and into USA, Europe, Middle East and Asia. I’ll spare them the public name-check, but the pattern is unmistakable: these aren’t companies waiting to be disrupted — they’re trying to be the disruption.

The cloud-communications platforms — the CPaaS world I’ve spent years in — surprised me most. At first this looked like the easiest room to work but while innovation is crucial the sector has turned hard toward cash-flow discipline; the instinct is to cut external commitments, not add them. And the category is rewriting itself in real time. Infobip now calls itself an “AI-first” cloud communications platform, dropping the CPaaS label altogether. Twilio has pulled off one of the genuine comebacks of the cycle — back to 20% growth and a billion dollars of annual free cash flow under Khozema Shipchandler, and openly repositioned as an AI orchestration layer rather than a bundle of APIs. Sinch has just changed its chief executive, the new direction aimed squarely at intelligent engagement and “the AI economy.” Three companies, one direction of travel. These platforms aren’t standing still — they’re spending inward, and betting the house on AI.

The hyperscalers and AI giants had the strongest presence in every room. They’re not short of capital or conviction — each has its own multi-billion-dollar venture arm. But a venture arm isn’t the same as a neutral, curated channel into the rest of the ecosystem, and that’s where a consortium like this earns its place: not as a substitute for what they already do, but as a way to augment it — extending their reach into the operators and platforms they want to work with, and adding more than one set of eyes to the diligence. When Nvidia puts a billion dollars straight into Nokia, that’s not a financial statement; it’s a directional one. The sharpest players will want both — their own bets, and a seat at a shared table.

And trust and identity — sovereignty, silent authentication, identity as a network service — which a year ago was a back-bucket afterthought and is now a front-line theme. The regulatory wave everyone treats as a burden is, for the players positioned right, the opening.

What changed

Here’s the part that actually shifted over the quarter.

I went out thinking the hard problem was one of mapping — figure out who’s who, match the innovation to the need. I came back convinced the market is already describing, out loud and unprompted, the exact problem worth solving. It kept saying three things.

First: there is no third door. Every midsize communications company faces the same forced choice — acquire your way up the stack, or make yourself attractive enough to be acquired. Seat-based pricing is giving way to consumption; adjacencies are converging; the leaders spend billions to consolidate and the laggards go private. Salesforce absorbing Fin, NiCE taking Cognigy, a billion and a half flowing into Genesys — the evidence stacked up all quarter. Standing still is not on the menu.

Second: AI is a reset, and the value is moving to orchestration. Not to the models — those are commoditising fast — but to the layer above them: model selection, where inference actually runs, how it’s governed, where the data is allowed to live. Orchestration, sovereignty, the control plane. A year ago that was a niche architectural debate three people cared about. This quarter it was the single biggest thing in the room, everywhere I went. Whoever owns that layer owns the margin.

Third — and this is the one that matters most — the people backing innovation directly keep getting stranded. An operator makes a direct bet; the strategic rationale evaporates a year later; the position sits orphaned on the balance sheet. A startup ties itself to a single carrier and is one reorganisation away from irrelevance. Almost everyone I spoke to carried a version of this scar. The problem isn’t a shortage of innovation, and it isn’t a shortage of capital. It’s that the two keep meeting in ways that leave both sides exposed.

You don’t arrive at that picture from a desk. You get to it from a hundred conversations.

Where this goes

So the quarter did its job. It wasn’t a victory lap and it wasn’t a pitch tour — it was a quarter of engaging and learning, and of taking some serious, concrete steps toward a model that answers the problem the industry keeps describing for itself. For Sandbox and CPaaSAA both, that’s the real result: not a stack of meetings, but a much sharper sense of what we’re building and who it’s for.

It wasn’t only learning, either. Some of the quarter’s most significant conversations were with the big players — a standout meeting with Google in Copenhagen, a great one with Cisco in New York, real time with AWS. And then the clearest signal of all: our first partner formally came on board to back what we’re building, and within days opened the door to Nvidia. The big guys are paying attention — and when the people who’ve just said yes start making your introductions, you know a thing has momentum of its own.

That momentum is now getting some structure — and it’s the part I’m most excited about. Here’s what actually struck me over these past weeks: it wasn’t any single meeting, it was how many genuinely senior people — across operators, hyperscalers, platforms and the standards world — offered to help without being asked. When that many serious people independently lean in, the move isn’t to thank them one by one. It’s to put them in the same room.

So that’s the next step. The dinner in Copenhagen was the prototype; Leonard Lee of neXt Curve, the analyst around the table, wrote on LinkedIn afterwards that it was the nexus of the network-API world — and the phrase landed, because everyone there had felt the same thing. That’s exactly what this should be. A nexus — a small, deliberate, frankly exclusive group of the people most committed to pushing this forward. The ones who’ll make the introductions, compare notes candidly, shape the pillars that matter — where intelligence runs, how trust and identity get solved, what the orchestration layer really becomes — and help turn a shared diagnosis into something built. Not a panel. Not a mailing list. The room where this actually gets worked out — and, if these past weeks are any guide, the logical next step for anyone who’d rather shape where this industry goes than watch it happen. I’ll share the specifics in a few weeks. If you’re reading this and feel you belong in that room, you probably already know it — let’s talk.

And there’s no shortage of room to do the work, the journey continues. The RAISE AI Summit in July, BATIC in August, the Alianza Summit and CPaaSAA Summit (CASA26) in September, MWC Doha in November — a runway across three or four continents to keep the conversations moving. BATIC is where we’ll lay out the “no third door” argument and the full market read in public for the first time. CASA26 in Amsterdam is where we put the rest of it on the table: what we’ve learned, who’s gathered around it, and the shape of the answer.

A quarter ago this was a set of questions. It’s becoming a direction.

And the road, it turns out, was the only way to find it.