In September I published State of BIM 2026, a market briefing for people building, funding and buying AEC software. It ended with five things to watch over the next 18 months. Each came with a test, the specific event that would show it had moved.
The briefing is no longer available, so here are the five as I set them out in September, trimmed for the web. In my next post I'll score each one against what has happened since.
1. A BIM 2.0 tool reaches LOD 350 on a real project
LOD 350 is the level of detail where wall assemblies are fully specified, services are coordinated and the model can support construction. It's where a BIM 2.0 platform becomes a real alternative to Revit on a typical architectural project. Below it, the new tools are useful additions to a Revit workflow. Above it, switching becomes a practical question.
Qonic has the most credible route there. It's IFC-native, built for detailed geometry from the start, and its Components release in August was the first significant step. Snaptrude's roadmap targets the same level of detail, after eight years of development.
A reference project is necessary but not sufficient. One project can be managed. The edge cases the incumbents have worked through over the last few decades show up on the hundredth project, not the first.
The test: Qonic or Snaptrude publicly references a project delivered on its platform at LOD 350.
2. The major vendors respond to open IFC
That Open Company's components and the wider open-source IFC tooling are maturing faster than the major vendors' roadmaps seem to reflect. The incumbents can integrate with the open stack, compete with it or ignore it. Ignoring it is the default for most of their product teams, and it won't stay viable as open tools close the gap outside authoring.
Nemetschek is better placed than Autodesk here. Graphisoft and Vectorworks have worked with IFC for years, and Nemetschek relies less on proprietary formats to keep customers. Autodesk's IFC support has improved in recent Revit releases but is still uneven across its products.
The test: one of the major vendors makes a significant public commitment to IFC-first development, or to integrating with the open stack.
3. Someone prices AI correctly
Every major AEC platform is bundling AI into existing subscriptions. The first company to price AI in a way that reflects what it costs, and that customers can understand and accept, will have an advantage over those that discover the problem later under revenue pressure.
The structural move is active-user rather than named-seat pricing, with AI computation charged separately from access. Autodesk Flex shows enterprise AEC buyers can handle consumption pricing, but it covers access to existing products, not the cost of AI. A challenger with an AI-heavy product and less legacy pricing to defend is better placed to move first.
The test: one significant AEC product moves from per-seat bundled AI to active-user plus consumption pricing.
4. Someone holds the building's data after handover
Most BIM models are abandoned at handover. The building then operates for 50 years, and the model sits in an archive, drifting further from reality with every change.
That's starting to change. Scanning hardware that cost six figures five years ago now costs four. Drone capture of large buildings is routine, and AI has cut the time from point cloud to usable model. More owners in infrastructure, healthcare and education are writing twin-ready deliverables into their contracts. What's forming isn't a continuously optimising replica. It's a usable record of a building that supports maintenance, space planning and compliance.
Four kinds of company are converging on it. Design-tool vendors are extending into operations, with Autodesk Tandem the clearest example. Bentley's iTwin already operates at scale in infrastructure. FM platforms are moving towards BIM. Purpose-built lifecycle platforms are starting from the owner's needs. The winner will be decided less by architecture than by who becomes the owner's primary data relationship first.
The test: the first large contract, for a government estate, a health system or a big portfolio, that names a digital twin platform as its operational data layer for ten years or more.
5. An AI-native way of authoring buildings emerges
This is the most speculative of the five, and possibly the most consequential.
BIM 2.0 set out to rebuild the authoring environment in the cloud and the browser, but kept the same production model underneath: trained professionals building detailed parametric models from which the documentation is derived. If AI can handle the documentation, and Glyph and Swapp are showing it can at increasing scale, the authoring tool doesn't need to carry the same weight. It could be lighter, focused on design intent and constraints, with agents doing the production work.
On 8 September Motif made the clearest claim to this so far, launching Motif Design as an agent-native platform where designers and AI agents work on the same live model. It still streams from Revit and Rhino rather than replacing them. An announcement isn't a delivered architecture.
The test: Motif or a competitor demonstrates agent-native authoring on a real project, end to end, without the incumbent tools in the chain.
The useful question for anyone building, buying or funding AEC software is which of these five they're most exposed to.