AI Is Driving the Future of Real Estate Returns

It was one of many case studies tucked into a Berkeley course I took that left me with this nagging feeling that everything is about to shift. The company, RS Metrics, used satellite imagery to count cars in parking lots outside big-box stores across America. That data, nothing more than an overhead snapshot of asphalt and metal, turned out to be predictive of revenues. Hedge funds were actually using it to get a trading edge, because it accurately predicted stock movements before quarterly earning calls even happened. I thought that if something as ordinary as parked cars could offer this kind of insight, what else might data, machine learning and AI have in store for real estate?

As someone who co-founded a boutique (it's a nicer word than small after all) private equity real estate fund in Hong Kong, this wasn’t just curiosity. It was a signal. A glimpse into how computer vision and machine learning can turn unconventional data sources into valuable business insights and "see" value before we do.
And I couldn’t shake the thought that our industry's old tools, like site visits, spreadsheets and gut feeling might soon be relics from a time that moved more slowly.

The Algorithms Are Already Here

According to The Business Research Company, in 2024 alone, the AI-in-real-estate market grew by over 36%, hitting $222 billion globally. To put that in perspective, that’s more than double the size of the global PropTech market just five years ago, and it’s still growing. The Business Research Company projects a compound annual growth rate of 34.4% through 2033, pushing the market toward nearly $1 trillion by 2034. That kind of trajectory isn’t just impressive; it’s rare, even in high-growth sectors like fintech or green energy. And yet, despite the headline numbers, many traditional real estate firms are still only experimenting with AI tools, suggesting this market is just getting started.

Across both residential and commercial sectors, AI systems are now digesting everything from zoning documents, satellite images, tenant behavior patterns, and rent forecasts.

Automated Valuation Models (AVMs) are recalibrating in real-time as new data flows in. Firms like HouseCanary and reAlpha aren’t just promising faster analysis, they're offering smarter analysis, with median valuation error rates dropping to just 5%.

Where once we relied on market comps and experience, AI now ingests millions of data points, from foot traffic to climate risk, to make investment decisions. Tools like CanaryAI let brokers and investors ask natural-language questions about pricing trends or risk profiles, collapsing what once took days into seconds. And global players like JLL are beginning to bake generative AI directly into their platforms. Their proprietary model, JLL GPT, is designed to make commercial real estate data more accessible and actionable, surfacing insights that previously required manual digging. While still in the early stage, the implications for tasks like lease analysis, financial review, and forecasting are significant.

This isn't about hypothetical futures. It's already changing how capital is deployed, how risk is priced, and how opportunity is found. And while the transformation may seem tech-first, it’s as much about mindsets as it is about models.

Levelling the Field or Widening the Gap?

Here’s the irony: AI is both a threat and a gift to smaller players. On the one hand, funds like ours simply can't compete with the resources of mega-funds building in-house AI divisions. But on the other hand, the democratization of tools is real. You no longer need a data science team to access predictive insights. Platforms are emerging that allow nimble operators to plug into the same intelligence streams used by institutional giants.

As Andrew Ng put it, we should think of AI as electricity in the early industrial age: at first it powered factories, then it redefined how everyone lived and worked. Today, even without coding expertise, we can build workflows, simulate markets, and extract insights from satellite data in minutes. I have zero coding knowledge but managed to build an app for analyzing monthly cashflow, projected capital gains, net equity etc. for an investment property in Singapore in just an hour using Replit. It’s called “vibe coding”, where you describe what you want in plain language and let AI generate the code for you. The challenge isn't access; it's vision. Who sees the shift early enough to ride it?

In that sense, AI is becoming what "insider knowledge" used to be. It doesn't just level the playing field; it rewires it entirely. In Singapore, for instance, PropertyGuru sits atop a pile of proprietary data, and AI will only make that advantage stronger. These giants are widening their moats.

But the same AI tools that make big companies even harder to compete with are also giving smaller firms surprising ways to compete—and sometimes even win. A boutique agency with five people and a smart automation stack can now outmaneuver a legacy firm that's still clinging to spreadsheets and sales scripts. The difference? Execution, not scale.

So, are we witnessing a David vs. Goliath moment, or the cementing of digital empires? The answer may be both. AI is widening the gap between the adaptive and the complacent. What matters now isn’t just size—it’s how fast you can move. And while data-rich giants still have the edge, they often move slowly.

That’s the window for challengers. Some will outmaneuver incumbents by using off-the-shelf AI tools more creatively. Others will compete by building what the giants don’t have: high-quality, niche-specific data. In some cases, that means collecting it themselves, on the ground, by hand, or through clever partnerships. It’s not about matching the scale of a platform like PropertyGuru. It’s about creating just enough signal in a focused domain to make smarter, faster decisions. Speed matters. But so does originality in how you source and shape the data you feed into these tools.

Still, there’s an uncomfortable truth: without meaningful access to data, even the most agile firms may struggle to benefit from AI. The playing field is only as level as the information it's built on. In that sense, the true competitive divide may not be about who uses AI—but who owns the raw material that fuels it.

The real risk is standing still clinging to the belief that gut instinct alone will carry us through. What the parking lot story taught me is this: by the time something becomes obvious, the alpha is already gone.

Looking Forward

Let's be clear — AI isn't magic (yet). I've watched firms automate lead scoring only to see conversion rates tank because the model got obsessed with clicks while missing all the human subtlety that actually matters. The sweet spot is using AI to get smarter, not to stop thinking entirely. It's still people — sharp, adaptive, ethically aware people — who make these tools work. Go too far down the automation rabbit hole without any judgment, and you'll probably regret it. I don't think AI will replace human judgment entirely. Real estate, after all, involves plenty of emotion, relationships, negotiation, and vision. But I do believe it will reshape the pace, precision, and thinking behind our decisions.

For my real estate business, and for me personally, this moment feels like standing on a threshold. The tools are here. The data is speaking. The question is: who is listening?

And more provocatively: who will be left behind because they weren't?