The AI-Powered 'Secret Sauce' Fueling Oxford Properties Group's New Investments
One of North America’s largest pension-backed real estate companies has been privately developing artificial intelligence that quantifies the value of a property down to every tree, blade of grass and speck of dirt, in an effort to drive smarter investment decisions.
Oxford Properties Group, which has a $62B, 150M SF real estate portfolio, has created what it calls Project Alpha, an AI-powered technology that uses information from around 50 datasets to create models of properties in markets around the globe.
Oxford, which developed the technology in-house along with Australian portfolio company Investa Property Group, says Project Alpha is the first technology that quantifies this level of detail for not just a property itself, but the environment around it.
“There are clearly indicators that say, a piece of dirt at a particular location has a certain amount of value, regardless of what’s in it,” Oxford Properties Group Chief Operating Officer Dean Hopkins told Bisnow in an interview. “On top of that, there are certain things that you can do to that piece of dirt that make it worth more.”
Hopkins — who had spent 25 years working in the technology sector as a startup founder and a technology company executive as well as a technology adviser for Thomson Reuters before joining Oxford in 2019 — described the system as a mix of the type of market-level information the Bloomberg Terminal holds combined with the data science of a quantum model, but for property investment instead of the stock market.
Project Alpha places numerical value on aspects of a potential investment that would be otherwise opaque, by reconstructing digital twins of the environment surrounding buildings, in addition to twins of the structures themselves.
“We're building algorithms and hunting for pieces of software that are using that data to inform our teams about where to look,” Hopkins said. “Now you don't have to hunt in the entire forest, we're going to tell you these very specific parts of the forest and give you real confidence and conviction.”
Project Alpha's technology has propelled the company's life sciences venture, which it calls "Project Warp Speed" internally. Oxford has acquired around 1.2M SF in life sciences investments across six deals in the United States this year so far, a spokesperson said.
Project Alpha's technology was used to guide the purchase of a 136K SF office building at 1930 Boren Ave. in Seattle for $119.1M, which is set to be converted to a life sciences use.
“People were like, ‘Huh. A life sciences deal in Seattle. That's kind of interesting,’" he said. “And it's because when we modeled it, we said, 'Oh, something really special is going on in Seattle, and we can tell exactly where in Seattle and why in Seattle.'”
The “secret sauce,” as Hopkins described it, that led Oxford to make the deal is the technology evaluates every inch of a building’s land and surroundings in order to aid in investment decisions, Hopkins said.
Hopkins and Investa General Manager for Innovation Joanna Marsh helm a team of a dozen data scientists, data engineers and programmers behind Project Alpha who are working to perfect that variable through feedback from investment and leasing teams at Oxford and Investa, Marsh said.
“As we start aggregating this base intelligence layer, the data that underpins all these models, we start to see things we didn't even know that we didn't know, and we start to overlay the data,” said Marsh, who has spent over 15 years in commercial real estate.
The team began development of the project two years ago, after a major turning point for Oxford Properties. In 2018, it reshaped its ambitions in New York from a passive partner on numerous major development projects, including with Related on Hudson Yards, to active developer, The Real Deal reported at the time. It acquired and redeveloped St. John’s Terminal near the HighLine in SoHo and landed Google as the building’s anchor.
Project Alpha is the kind of technology that could only be created by real estate experts — those that understand the nuances of qualitative value in real estate and are able to ensure the quantitative data grounding the technology remains true to that, Marsh said.
“There's such subtlety and sophistication and opaqueness, and it's very tricky to understand how real estate works, how the models work. No one's really modeled it before,” Marsh said. “When you go out right now into the property markets, a lot of the data sets, both the core and non-core, just lack integrity.”
“They are using the digital twin tech to layer in parts of the data landscape that others perhaps have not built to date and I think that’s really, really unique and interesting,” Beckerman said. “I think it’s going to give them a real competitive advantage, it has tremendous potential in the marketplace, I think it’s a brilliant move on Oxford’s part.”
Increasingly, large real estate companies like Oxford Properties are building their own property technology system because what they need is not on the market yet, Beckerman said.
“Increasingly, we’re seeing more and more real estate companies deciding to build versus buy, and that’s a healthy and natural evolution of the real estate tech ecosystem,” he said. “The end result is that companies building their own proprietary solutions will be able to use that technology as a real competitive market advantage versus their competitors, who are perhaps more focused on off-the-shelf solutions.”