Anthony. I like your insights because you clearly have deep knowledge of the sector. Increasingly though, it takes time to follow through your in depth analysis through to useable conclusions. So I use AI to do this! - Maybe that is food for thought for another article -
Here are the core points in plain English:
• Five players now compete in CRE tech, not two. AI labs (OpenAI, Anthropic), large enterprise consortia (backed by Blackstone, Goldman etc.), established data incumbents (Yardi, CoStar, Argus), specialist startups, and operators who build their own tools. Each occupies different ground.
• Generic PropTech is being squeezed out. Any software that is essentially a wrapper around a general AI model — chatbots, reporting overlays, generic workflow tools — will simply be replaced by the AI labs directly. The configurable middle tier is being commoditised.
• A VC framework offers three useful tests for whether a product/service survives. Is the work multi-step and complex? Are you the system clients run work through (not just a plug-in on top of something else)? Are you judged on client outcomes rather than how impressive the technology looks? Anything that fails those tests is vulnerable.
• Established incumbents still hold strong positions but can no longer coast. For 15 years Yardi, Argus etc. faced no real competitive pressure. Challenger routes now have viable AI-driven economics for the first time. They will probably still win this cycle, but they have to actually compete now.
• The viable startup owns one painful process end to end. Not selling AI tools into the industry, but becoming the place a single regulated, multi-step process actually happens — and getting better at it with every client.
• Operators’ instinct to “just buy software” is becoming outdated. Building in-house was not viable for mid-tier firms two years ago. It increasingly is. Any vendor commitment today is a bet on whether your supplier improves faster than you could build it yourself.
• Consortium deployments quietly commoditise operators’ proprietary knowledge. Engineers embedded across multiple firms observe how each one works, then fold that into a standardised product sold to the whole industry. Your edge gradually becomes everyone’s baseline.
• The central question is which of the five players your firm actually is — and whether you are competing on ground you can defend. Most firms are not asking it clearly enough, and by the time they do, the strong positions will already be taken.
Anthony. I like your insights because you clearly have deep knowledge of the sector. Increasingly though, it takes time to follow through your in depth analysis through to useable conclusions. So I use AI to do this! - Maybe that is food for thought for another article -
Here are the core points in plain English:
• Five players now compete in CRE tech, not two. AI labs (OpenAI, Anthropic), large enterprise consortia (backed by Blackstone, Goldman etc.), established data incumbents (Yardi, CoStar, Argus), specialist startups, and operators who build their own tools. Each occupies different ground.
• Generic PropTech is being squeezed out. Any software that is essentially a wrapper around a general AI model — chatbots, reporting overlays, generic workflow tools — will simply be replaced by the AI labs directly. The configurable middle tier is being commoditised.
• A VC framework offers three useful tests for whether a product/service survives. Is the work multi-step and complex? Are you the system clients run work through (not just a plug-in on top of something else)? Are you judged on client outcomes rather than how impressive the technology looks? Anything that fails those tests is vulnerable.
• Established incumbents still hold strong positions but can no longer coast. For 15 years Yardi, Argus etc. faced no real competitive pressure. Challenger routes now have viable AI-driven economics for the first time. They will probably still win this cycle, but they have to actually compete now.
• The viable startup owns one painful process end to end. Not selling AI tools into the industry, but becoming the place a single regulated, multi-step process actually happens — and getting better at it with every client.
• Operators’ instinct to “just buy software” is becoming outdated. Building in-house was not viable for mid-tier firms two years ago. It increasingly is. Any vendor commitment today is a bet on whether your supplier improves faster than you could build it yourself.
• Consortium deployments quietly commoditise operators’ proprietary knowledge. Engineers embedded across multiple firms observe how each one works, then fold that into a standardised product sold to the whole industry. Your edge gradually becomes everyone’s baseline.
• The central question is which of the five players your firm actually is — and whether you are competing on ground you can defend. Most firms are not asking it clearly enough, and by the time they do, the strong positions will already be taken.
Thanks Peter
I said all that in 1600 words!
Good summary.