There's a narrative circulating in AEC tech circles that the major BIM players are in trouble. Autodesk's stock is down significantly from its highs. Nemetschek and Trimble have had difficult quarters on the market. And the BIM 2.0 challengers - the browser-native, database-centric platforms that were going to disrupt the incumbents - haven't yet delivered on that promise.
These are two separate stories being conflated into one. The stock pullback is a valuation story, not a business story. And BIM 2.0's slower-than-expected progress has nothing to do with AI - it reflects the genuine engineering complexity of rebuilding production-grade BIM tooling from scratch. AI is a different force in this market - one I've written about separately in The AEC Software Moat Map - and conflating it with the BIM 2.0 narrative obscures both.
I've been close enough to this industry for long enough to find the conflated version frustrating - not because it's entirely wrong, but because it's asking the wrong questions.
The Autodesk Number Is a Valuation Story, Not a Business Story
Autodesk's stock trading 30% below its 52-week highs does not mean Autodesk's business is deteriorating. Their AECO segment continues to post double-digit revenue growth. Margins are expanding.
What happened is simpler: software valuations got stretched to extremes during the cloud growth era, and now the market is repricing to something more realistic. This is normal. It's happened before - to Salesforce, to Workday, to ServiceNow - and in every case the business continued to perform while the stock found a more honest level. The business is fine. The share price just had further to fall.
The same applies to Nemetschek and Trimble. Nemetschek's subscription transition is real and progressing. Trimble's recurring revenue mix is improving. Neither company is in crisis. Their stocks were simply priced for a version of their future that required everything to go perfectly - and markets eventually stop believing that.
If you're building or investing in AEC tech and reading Autodesk's chart as a signal that the platform is weakening, you're reading the wrong signal. The incumbents have real problems, but they're not the problems the stock price suggests.
BIM 2.0 Is Caught in the Middle
The BIM 2.0 story in 2026 is not failure. But it is an uncomfortable position.
On one side, the BIM 1.0 incumbents haven't weakened the way the challenger narrative assumed they would - as covered above, their businesses are fundamentally sound and the engineering depth they've accumulated is harder to replicate than the hype cycle suggested. On the other side, AI is enabling a different kind of shift that wasn't part of the original BIM 2.0 thesis at all - one that may leapfrog incremental cloud-native improvement entirely. BIM 2.0 was conceived in a pre-AI world. It's now building toward a market that AI is rewriting from underneath.
I've spent enough time building design software to know how complex BIM tools actually are. The jump from "browser-native collaboration concept" to "product that architectural firms will trust with live project data" is not primarily a UI or infrastructure problem. It's a depth problem. BIM tools have fifteen years of hard-won workflow logic embedded in them - phase management, annotation behaviour, parametric relationships, coordination protocols. Replicating that depth from scratch, cloud-native, while simultaneously making the product good enough to sell, is a genuinely difficult thing to do.
The depth problem is real, and none of these teams have yet reached full authoring parity with the incumbents. But that doesn't mean they've stalled. The more accurate story is that most of them are now pivoting toward AI-assisted workflows - targeting the highest-friction edges of the design process rather than attempting to replicate Revit's full workflow depth in one go. That's a sensible adaptation - and arguably a signal that the more interesting question isn't whether BIM 2.0 displaces BIM 1.0, but what an AI-native approach to the whole problem might look like. That's the subject of Part Two.
AI Is Real, But Incremental
AI in AEC in 2026 is not future speculation. It's in workflows today - clash detection, performance analysis, documentation automation, coordination checks. Substantial parts of the industry are using it actively, and the genuine impact is real: faster outputs, better coordination, less grunt work absorbed by skilled people who should be doing harder things.
But it's worth being precise about what Revit and ArchiCAD actually do - because AI's impact is not evenly distributed across it.
Core parametric modelling - phase management, annotation logic, structural grid relationships, coordination protocols - is still largely untouched by AI. These haven't been rewritten; they've been accelerated at the edges. That's meaningful, but it's not transformation.
Documentation is a different story. A significant portion of what BIM tools are used for every day isn't modelling at all - it's producing construction documents: views, sheets, tags, dimensions, schedules, drawing sets. And this is where AI is moving considerably faster. Glyph automates view creation, dimensioning, tagging, and sheet packing inside Revit, cutting what previously took half a day to minutes. Swapp goes further, automating full DD/CD sets by learning a firm's office standards from historical projects and applying them to new work. At the earlier design stage, Finch3D handles generative floor plan layout with direct two-way export to Revit - compressing the concept-to-documentation pipeline from the other end.
These aren't experimental tools. They're in active use. The documentation layer - which has always been where much of the production time sits - is being automated faster than the modelling layer, and largely within the incumbents' own platforms rather than outside them.
The mistake is to project this trajectory uniformly and conclude that core BIM will be fully AI-automated by next year. The modelling depth remains hard to replicate. But the documentation story is already a different one, and it's moving quickly.
What This Means If You're Building
The incumbents' real vulnerability isn't their share price or their scale. It's their ability to adapt their existing products - which were built on different technical assumptions - to a market that increasingly expects AI-native behaviour, real-time collaboration, and open data access.
The companies that navigate this transition well will be the ones that invest in their platform's AI layer while it's still their choice to do so, rather than being forced to retrofit it later. That requires a clear view of which parts of your product are genuinely defensible - and which parts you're assuming are defensible because they've always been.
I'll cover the forward-looking side - digital twins, point solutions, and what BIM 3.0 actually looks like - in Part Two.
If you're building or scaling an AEC tech business and thinking through where BIM fits in your strategy, I'd welcome the conversation.