State of BIM 2026 - Part Two: Where it goes from here
In Part One I wrote about the BIM incumbents - why the stock pullbacks aren't the story, and why the BIM 2.0 execution reality was entirely predictable to anyone who's built in this space. This is the forward-looking half: digital twins, point solutions, and what I think BIM 3.0 actually means.
Point Solutions Are Thriving for a Structural Reason
The explosion of specialist AEC tools - automated clash detection, generative components, scan processing, carbon calculation - looks from the outside like fragmentation. From the inside, it makes perfect sense.
The major platforms have consistently underinvested in the narrow, specialist workflows that define how specific disciplines actually work. A structural engineer's analysis workflow, a MEP coordinator's clash resolution process, a sustainability consultant's carbon reporting - these are genuinely complex, discipline-specific, and not where Autodesk or Trimble focus their core engineering effort. That's a gap, and gaps attract products.
Point solutions that work well tend to do a few things consistently. They charge on project economics rather than recurring subscriptions - which aligns their revenue with how AEC firms actually work, project to project. They're adopted in narrow contexts where the incumbent doesn't compete. And they survive because being genuinely best-in-class at one job is a viable business model in an industry where the alternative is a massive platform that does everything adequately.
The interesting strategic question for anyone building a point solution is whether the niche can support a standalone business, or whether it eventually needs to integrate into a broader platform to reach full value. Both models work. But they require fundamentally different commercial strategies from day one.
Digital Twins Have Moved Past the Buzzword Phase
A few years ago, "digital twin" meant something between a marketing term and a research project. In 2026, it's a real market segment with real procurement happening.
What's actually driving this is less sophisticated than the utopian version. Cheaper scanning hardware and AI-assisted scan-to-BIM processing have made asset capture economically viable for a much broader set of buildings and owners. Revit and IFC models are increasingly being retained post-handover for operational use rather than being archived after construction. Owners - particularly in infrastructure, healthcare, and education - are demanding twin-ready deliverables as a condition of contract.
The result is a genuine market forming around lifecycle data management: performance modelling, sustainability tracking, integrated asset management, space planning. The question isn't whether buildings will be managed digitally. It's who owns the platform that makes that practical, and what data standards become the common layer across it.
This is also where the most interesting long-term competitive dynamics sit. The firm that owns the lifecycle data layer - the operational twin that persists across the building's life - has a different relationship with the owner than the design tool that produced the original model.
Where This Could Go? BIM 3.0?
Having spent years building design tooling, the logical extension of AI integration points somewhere interesting. Not a prediction - but a direction that feels increasingly hard to argue with.
The documentation layer is the tell. As covered in Part One, tools like Glyph, Swapp, and Finch3D are already automating significant portions of what BIM platforms are used for every day - not the modelling, but the production of drawings, sheets, tags, schedules, and full documentation sets. This is happening now, within the incumbents' own platforms. The constraint that has always defined BIM workflows - that authoring and documentation are inseparable, and both require platform-trained professionals - is beginning to break.
Which points to something structurally important: a new lightweight design tool, paired with AI-handled documentation, is a different proposition to anything BIM 2.0 built toward. BIM 2.0 started in a pre-AI world and set out to rebuild the BIM 1.0 stack from scratch - cloud-native, browser-based, database-centric. The ambition was right. But it carried the same underlying assumption: that the authoring environment is still where the work lives, just delivered differently. The documentation burden stayed.
BIM 3.0 - if the label means anything - starts from a different question entirely. Not "how do we rebuild the authoring environment?" but "if AI can handle the production layer, what does the design tool actually need to do?" A world where modelling grunt work is progressively automated. Where lightweight front-ends let more of the project team engage with the model directly, without needing to be Revit-trained. Where documentation engines generate drawings from model state rather than requiring manual production effort. Where information - compliance, cost, carbon, programme - becomes primary, and geometric representation becomes the output rather than the work.
That's not an evolution of BIM 2.0. It's a different starting assumption.
If you follow the naming convention, you might call it BIM 3.0. The label doesn't matter much. What matters is that three or four years ago this direction looked academic. Now it looks like a question of timing.
But it raises real questions that nobody has clean answers to yet. Can existing platforms adapt their architecture fast enough, or does information-centric production require building from a different foundation? Will new tooling displace legacy stacks, or integrate alongside them? How much do workflow ecosystems and switching costs protect the incumbents as the underlying production model shifts?
These are the questions worth watching - not because they're unanswerable, but because who answers them first will matter a great deal.
What I'd Watch in the Next 18 Months
A few specific things I think are worth tracking:
Who captures the lifecycle data relationship. The post-handover digital twin market is forming now. The companies that establish themselves as the operational data layer - not just the design tool - are building a very different kind of business. This is an undercontested space.
Whether any BIM 2.0 challenger breaks through on detailed authorship. Two teams stand out as front runners. Qonic has attacked detailed modelling from the start - their architecture was designed for it, not retrofitted onto a concept tool. Snaptrude has been building in this space for over eight years, and their roadmap now explicitly extends to LOD 300-350. If either delivers genuine parity with Revit for detailed design on a real project, the transition timeline compresses materially.
How the majors respond to That Open Company's ecosystem. The open-source IFC stack is maturing fast. If Autodesk, Nemetschek, and Trimble don't have clear strategies for how they integrate with - or differentiate from - the open stack, their position in the non-authoring BIM space weakens faster than their roadmaps probably reflect.
The first AEC software company to price AI correctly. Most are bundling AI features into existing tiers right now. The first company to build a pricing model that reflects AI's actual cost structure - without making it feel complicated to customers - is ahead of its competitors in the commercial model conversation that the whole industry will eventually need to have.
If you're working through what any of this means for your AEC technology business, I'm always up for the conversation.