Last week, I attended the first PTC Next 2026 event in Chicago. The event is part of a new PTC NEXT Innovation Seasons events.
First, a thank you to PTC for inviting me to learn about the company’s new strategy and its latest products. It was two days extremely well spent, and over the next few Beyond PLM blogs I want to share my perspective on what I learned.
PTC unveiled a wave of new products. Here is a quote from the press release: “PTC introduces PTC Orbit and PTC Jetstream, a new AI platform, 12 AI agents, 10 new integrations, and 100+ enhancements across CAD, PLM, ALM, and SLM solutions. PTC delivers the Intelligence Layer across the portfolio, giving manufacturers a stronger foundation for product data, deeper AI capabilities, and a more connected Intelligent Product Lifecycle”
One disclaimer up front: my blog is not news reporting. If you want the official version, the presentations and announcements are available online (PTC Next on Demand), and I’d encourage you to navigate to PTC’s own materials and follow the company directly.
What I offer here is something different. It’s my perspective, seen through the prism of more than a decade of writing about, and learning from, the transformation of PLM.
I had a conversation in Chicago that I keep coming back to. I was standing in front of a PTC representative walking me through the “Intelligent Product Lifecycle,” the framing at the center of PTC’s new strategy. It was a very good pitch: clear, confident, well-rehearsed. And at some point I said the thing that twenty years of covering PLM has trained me to say – PLM Holy Grail. Vendors have promised “we connect everything” for a very long time, and the hard problems (adoption, heterogeneous stacks, implementation cost, the ROI question that never quite goes away) are all still here. So I asked the obvious question: when every PLM vendor now calls itself “intelligent,” is that a real architectural difference, or just a new label?
Then I reached for an old marketing story. In the 1920s, Lucky Strike ran a famous campaign: “It’s toasted.” The line implied something special. The catch is that every cigarette is toasted; that’s how tobacco is processed. The slogan didn’t describe a difference. It described a process everyone already used and dressed it up as an advantage.
So when the rep finished, I asked the version that had been forming in my head the whole time: is “intelligent” a real difference, or is it toasted?
Most of what carried the “intelligent” label at the event was a reframing of things PLM vendors have described for years. However, one announcement was different in kind rather than degree. PTC introduced a new, cloud-native collaboration platform (PTC Jetstream) that holds product data and, more importantly, the decisions made about it, and it is a new architecture: cloud and multi-tenant, built deliberately outside the tradition and existing system of record. That architectural choice is worth more attention than the product category suggests, and I’ll come back to it in my next articles.
I want to be fair about my own position here. I didn’t come to PTC Next to dunk on PTC. I came genuinely looking for an answer, because this is a question I’ve been asking for a very long time. And because, as it turned out, PTC’s own leadership spent two days on stage making arguments I happen to agree with. The interesting tension of the whole event is that the diagnosis was often right. The open question is whether the architecture behind it is.
I’ve Been Asking This Since 2011
The reason “intelligent” doesn’t automatically impress me is that I’ve watched the industry promise product intelligence before, under different names.
Back in 2011, I wrote about the role of business intelligence in PLM. My conclusion then was that BI was addressing exactly the right problem (decision support, turning product data into something a human could act on), but with the wrong, reporting-era architecture. It was bolt-on analytics sitting beside the system of record, expensive and detached from the moment a decision actually gets made.
Fifteen years later, a rep is selling me “intelligent” PLM. The destination is identical: better decisions, made from product data. The question I had in 2011 survives completely intact. The problem is right. Is the architecture finally different this time, or is this BI déjà vu with a large language model where the reporting dashboard used to be?
That is the question this series is built around. In 2026, everyone will say “intelligent.” Siemens will say it. Dassault will say it. SAP will say it. A word that every vendor uses describes none of them. So across the next few articles I want to test one thing: what would have to be architecturally true for “intelligent” PLM to mean something, and did PTC Next show evidence of that, or just the adjective?
“Intelligent” as a Rename
Start with the case against PTC, because it’s the easy one to make.
On the surface, PTC Next looked like a rename. The event introduced “Intelligent Solutions” as a pillar, announced a count of new AI agents across the portfolio, and used a recurring “Product Innovation Wayfinder” honeycomb graphic that filled in, hexagon by hexagon, as each launch was revealed. It was a well-produced repackaging of the existing portfolio (Creo, Windchill, Codebeamer, Onshape, ServiceMax) under an AI banner.
If you stopped there, you’d have your answer. This is what a rename looks like: the same connected-lifecycle promise the industry has made for two decades, now narrated with newer vocabulary. Every PLM commentator writing from the press release will produce some version of that recap.
But I wasn’t writing from the press release. I was in the room for the keynotes, and the keynotes complicated the easy story.
How Neil Barua and Jon Stevenson Frame the Strategy
The most useful thing I can do, as someone who was actually there, is tell you what PTC’s two most senior people actually argued, because in several places they argued my own positions back to me from the main stage.
Neil Barua: The Diagnosis I Recognize
Neil Barua, PTC’s CEO, opened by framing this as the most transformative moment in the company’s forty years. Fine; every CEO says that. What caught my attention was how he defined the Intelligent Product Lifecycle. Strip away the staging and it comes down to this: connect product data on top of the systems of record, then apply AI, then remove the friction between lifecycle stages. Three pillars: connected through openness, powered by product data and fueled by AI, accelerated by SaaS.
I’ve heard that structure before, because I wrote it. In 2022 I described PLM’s path as a move from system of records, to system of engagement, to system of intelligence. Barua’s “data on top of systems of record, then AI” is that thesis, four years later, delivered from a PTC stage. I’ll take the validation.
I’ll also give him real credit for candor, which is rare at vendor keynotes. He admitted that no one really knows what this technology can do consistently at scale. He named the risks that are specific to manufacturing (precision, manufacturability, traceability, IP protection) and said plainly that AI is not naturally good at those things. And he set a principle I liked: “We are not in service of AI. AI is in service of us,” with no obligation to any one model provider. That last point quietly matches something I now even put in my own disclosures: keep the value in connected data and context, and treat the model underneath as replaceable.
So far, so good. But here’s where my 2022 article also raised a warning I have to apply now. The whole point of a system of intelligence is that it has to be native to the data layer, not a separate analytical thing bolted beside it; otherwise you’re back in 2011. “Data on top of systems of record” is precisely the phrase that decides which one this is. On top of, or part of? That distinction is the difference between architecture and toast.
Stephen Olive: Build the Thread First
The strongest line of the event, for my money, came not from PTC but from a customer. In a fireside chat, Stephen Olive of the US Department of Energy described running his sixth digital transformation. His framing was blunt: you don’t get paid to design, you get paid to ship. And his sequencing advice was sharper still: build the digital thread first, then layer AI on top of it. Do not use AI to create the thread. The real barriers, he said, aren’t tools or integrations; they’re the long-relied-on processes nobody wants to touch.
That is, almost word for word, the argument I made in task re-engineering before AI. My point there was that AI in PLM isn’t embedded intelligence so much as delegated work. An agent is a junior assistant you hand a bounded, clearly described task. And it only works if the work is explicit. Olive’s “thread first, then AI” is the same idea seen from the data side: you can’t delegate reasoning over a foundation that doesn’t exist yet. When a customer on PTC’s own stage says the hard part is the process and the thread, not the AI, that should tell you where the real work still sits.
Jon Stevenson: Openness, the Foundation, and the Master-Data Question
Jon Stevenson, the Chief Product Officer, made the boldest claim of the event, that PTC is the only company covering the complete lifecycle, and then conceded the load-bearing point underneath all the AI talk. His words: AI is only as strong as the data it can reason over, and siloed data produces fast-but-wrong answers. The answer, he said, is the product data foundation, connected by openness, with no rip-and-replace, so customers modernize at their own pace.
I agree with the premise so completely that I have to push on the conclusion. The “openness, no rip-and-replace” framing is an endorsement of an argument I’ve been making against the industry’s instincts: that the monolithic, single-database PLM dream is breaking down. You cannot pour the ocean into a swimming pool, and products today are built across suppliers, partners, and systems no single vendor owns. So when Stevenson says “openness,” I want to know whether the product data foundation is genuinely federated, or whether “foundation” is the single-source-of-truth claim wearing a newer, friendlier word.
And there’s a second question hiding in “product data foundation.” Staking a claim to be the foundation is a move in the master-data war I mapped in 2024, the long contest between PLM, ERP, CRM, and MES over who owns the product master. Windchill, Codebeamer, Creo, and ServiceMax as the backbone, with the new Orbit app aggregating asset-centric data across PLM, IoT, and service, is PTC positioning itself to own that layer. Fair enough. But owning the foundation answers what the data is. It does not, by itself, answer why a decision was made, and that gap is the whole game.
The agent strategy sits on top of all this: a count of agents already shipped, an Advise → Assist → Automate framework, “shipping today, not a roadmap.” I believe the shipping part. But agents are delegated workers, and delegated workers need both an explicit task and a memory of what was decided before. PTC has the agents. The question I carried out of those keynotes was whether anything underneath them actually captures the reasoning they’d need to reason over.
The One Honest Signal
If the keynotes had stopped at openness and agents, I’d have called it a sophisticated rename and moved on. One session stopped me.
Joseph June, who leads AI strategy, gave the most intellectually honest talk of the event. His argument: the models themselves are commoditizing. He openly questioned whether most of the audience could even tell the latest frontier models apart, and said it mostly doesn’t matter to them. The scarce resource isn’t the model. It’s product data prepared so that AI can actually reason across it, stored (in his framing) so it can be reasoned over, not merely displayed. He used an iceberg: the visible blockers (models, frameworks, use cases) hide the real one, which is that the data was never prepared for reasoning in the first place.
That is not a talking point I have to debunk. It’s PTC’s own AI lead validating the premise I’ve been building toward for years: that data and captured context matter more than generation. It’s the seed of an argument that there might be real architecture under the adjective. It’s also the moment the event got most interesting, because it implicitly admits the foundation isn’t done.
The Booth, Again: Where the Architecture Runs Out
Which brings me back to the booth, and the moment the whole thing crystallized.
When I pushed the rep (could I point this thing at my company’s data and get back “here are your problems”?), he didn’t bluff. He acknowledged that’s a different category from what exists today, and that the diagnostic vision becomes possible only once all the data is captured in connected repositories. In other words: the value depends on a layer that isn’t built yet. A PTC representative, unscripted, told me that “intelligent” is a promise contingent on capturing context that today mostly isn’t captured.
That sentence is where fifteen years of my own writing lands in a single point. In 2011, the right problem had the wrong architecture. In 2022, I argued the intelligence had to be native to the data, not bolted beside it. By 2025, I was writing that AI needs explicit work and a memory of the why behind decisions to be useful at all. PTC named the destination on stage: connected data, fueled by AI, reasoning across the lifecycle. The rep admitted in the booth that the road to it, the captured reasoning and decision history, isn’t paved.
I’ll note one more thing fairly. This is genuinely the direction PTC is moving, toward the new intelligence layer and service-history territory that, in my view, much of the industry has left undefended. The recursive-intelligence pitch around ServiceMax and the traceability examples the rep gave (knowing not just that a quality step happened but who did it and when, and why a design choice was made) are exactly the right targets. PTC sees the layer. The question is whether they’re building it or describing it.
What Is My Conclusion?
A rename markets the old promise with a new word. A new architecture would require capturing the reasoning behind product decisions in a form AI can actually use, and that is a far harder thing than shipping agents or aggregating a foundation.
PTC Next 2026 showed me something more interesting than a rename and less finished than an architecture. PTC’s leadership has, impressively, adopted the diagnosis, much of it the same diagnosis I’ve published over the last decade and a half. What I didn’t see, and what the rep honestly conceded I wouldn’t yet find, is the captured context that would make “intelligent” true rather than toasted.
So I’m not convinced, but I’m not dismissive either. I’m exactly where the best skeptics should want to be: with a clear test and a willingness to be proven wrong. Over the next few articles I’ll apply that test to PTC’s actual pieces (the data foundation, the argument about text-to-CAD, what AI agents really expose about the missing layer) and follow the evidence where it goes.
When a vendor’s own diagnosis is this good, the only fair thing to do is hold them to it. That’s what the rest of this series will try to do.
What do you think: is “intelligent” PLM a new architecture, or is it toasted?
Just my thoughts…
Best, Oleg
Disclaimer: I’m the co-founder and CEO of OpenBOM, an AI-native collaborative digital thread platform connecting engineers and manufacturing teams. My opinion can be unintentionally biased.
