While the full impact of the pandemic has yet to be realized, commercial real estate faces new uncertainties, including questions about the AI boom’s longevity, the spending strength of the U.S. consumer, and debt sustainability. In response to increased competition for quality deals, commercial real estate firms are restructuring their operations, using diverse data sources, accessing new capital, and forming new partnerships.
During a panel at this week’s conference titled “The Sea Change in CRE Lending,” industry veterans dissected how today’s lending environment compares with past downturns and what it will take to get deal activity back to pre-pandemic levels.
“Every cycle is a little bit different, and every cycle is a little bit the same,” said Greg Wolkom, head of REIT finance group and real estate syndicated finance at Wells Fargo. After the initial COVID-19 shock, the market followed a predictable pattern: rising interest rates led to troubled loans and a steep drop in property acquisitions. But this time, one crucial factor changed the trajectory.
Unlike in the Great Financial Crisis, banks entered this downturn better capitalized and better protected. “A lot of banks had done senior mezzanine deals,” Wolkom said, placing riskier debt with private capital providers rather than keeping it on their own books. “Folks rushed to say, ‘I’m going to buy all these deals at discounts from banks.’ That didn’t materialize. You didn’t see a lot of the portfolio sales that you had seen [in the last cycle].”
Wolkom pointed to an unusual dynamic: Private credit funds are offering returns in the low-to-mid teens for debt investments. “If you’re an equity player, and you have both a debt and equity component, what do you have to get as a premium to take equity risk?” he asked. “I think a lot of folks have so far said, the premium of what I need is not there to take that equity risk, when I can take that credit risk.”
New funds, data centers
One of the most striking shifts is the blurring line between real estate and infrastructure investing. Wolkom noted that infrastructure funds have raised capital equal to what real estate funds have amassed, and these investors are now crossing into real estate territory.
“We’re seeing infra come into data centers,” he said, citing a recent major financing Wells Fargo arranged for Blackstone in the marina sector that was funded entirely by infrastructure investors. Because infrastructure investors typically accept lower returns in exchange for stable, long-term cash flows, their entry into real estate markets is intensifying competition.
Five years after the pandemic, the fate of urban cores remains unresolved. “We’re still in the early innings of coming out of that,” said Shaina Doar, senior advisor at McKinsey & Company. Some cities with diverse, mixed-use neighborhoods are seeing foot traffic return, but office occupancy lag behind. For economic development officials, the question is no longer whether to act, but how to make interventions financially feasible. Converting empty office towers to residential use has emerged as one potential solution, though Doar noted it’s “never going to be the whole solution.”
While downtown revitalization struggles to attract private investment, data centers fueled by artificial intelligence demand are drawing both capital and government attention. “The thing that does pencil, if you’re thinking about economic development, is all the data centers,” Doar said, although she notes that these facilities typically locate outside traditional urban centers. “In the same way that you’re seeing the money from infrastructure and real estate coming together, you’re seeing teams across various entities within the public sector having to come together, because that is how they are going to get those things done. They’re trying to get AI to be a part of their economy.”
Housing takes center stage
Perhaps the most politically charged issue is housing affordability, which Doar said has moved to a top-tier political priority: “[For] the election two days ago, it was top of the ticket in terms of what politicians are talking about, because it is top of mind for most residents and most citizens,” she said.
Unlike data centers, housing doesn’t usually pencil without intervention, she said. Yet economic development agencies that once delegated housing to other departments now find themselves on the front lines to increase the housing supply.
Wolkom asked whether the broader trend shows municipalities becoming more receptive to development and less susceptible to NIMBY opposition. Doar suggested the shift is happening, but unevenly. “You see a split of cities that are willing to do creative things and even put a little bit of public juice into it,” she said, whether through regulatory reform or financial incentives or fast-tracked approvals. Calgary earned a mention for its ambitious initiatives, Doar said, while Chicago has seen minimal action, with just one office-to-residential conversion completed.
Revamping underwriting models
As cities experiment with new approaches to development and lenders navigate an evolving market, a quieter revolution is underway in how the industry evaluates opportunities. Ryan Luby, managing director and head of strategy at Wells Fargo, said few practitioners in the public and private sectors were creatively combining what he calls traditional and nontraditional data sources to make better decisions.
Traditional data, Luby explained, includes the standard inputs that flow into underwriting models: rental rates per square foot, occupancy levels, and other fundamental metrics. Nontraditional data encompasses factors that indirectly influence real estate value.
“The classic one here is foot traffic and local activation,” Luby said. In his previous role at Wells Fargo, he worked on urban revitalization projects with Doar, and his team-built models comparing traditional underwriting approaches with models incorporating broader data sets. The results were dramatic: “We showed that you boost explanatory power on the order of 50 to 60 percent if you incorporate a wider set of variables.”
Luby cautioned that capturing this value at scale requires solving four fundamental problems: gaining access to nontraditional data, ensuring the data is high-quality and regularly updated, integrating disparate data sources so they work together, and running sophisticated analytics on the combined dataset.
“There are some use cases where those things can be done,” Luby acknowledged. But he’s seeing a problematic trend in lending and capital markets: “There’s shareholder pressure to jump to GenAI and AI-driven solutions.” But generative AI only addresses the analytics piece—the fourth challenge. “It’s completely useless on one and two, and even a little bit of three,” he said. “So you need to solve the access and the quality issues before you can unleash a bunch of this value.”
This enhanced data capability could enable improved underwriting, pricing, and return modeling using a much wider set of data deployed in real time at scale, Luby said. He also envisions creating a comprehensive, live inventory of undeveloped land parcels across the entire country, enriched with information about local zoning flexibility.
This “qualified parcel scoring” system could dramatically reduce risk in site selection and early-stage development—helping answer questions about which locations pencil before significant capital gets committed. It’s the kind of tool that could help cities and developers identify which downtown office buildings are truly viable for residential conversion, or where new housing projects have the best chance of success.
AI real estate management
Alex Wolkomir, partner at McKinsey & Company, said that how the industry uses AI could determine whether these technologies deliver real value or become expensive distractions. “If you don’t measure it, you can’t do anything with AI, because the data needs to be there,” Wolkomir said. “But we’re not getting to the place where it’s actually easy to get data or measure things that were not possible [before].”
Wolkomir argued that the industry is reaching an inflection moment where simply generating insights or building dashboards no longer qualifies as meaningful progress. “We’re now at the place where AI has to get to an outcome or an action.” The question isn’t whether the technology can identify patterns, but whether it actually improves building operations, generates better returns on capital, or streamlines critical processes. “How do you measure what your AI is actually doing?” he asked.
Wolkomir challenged the industry’s standard approach to implementing AI—what he called the “use case” mentality: “Use cases are actually the wrong way to do AI. If you’re doing use cases, you’re not going to get value.” The use case approach, he said, leads to scattered, disconnected AI implementations across an organization. “It’s actually about domains, families of use cases, and how do you reinvent full processes like lending, zoning, and entitlements.”
For commercial real estate, this suggests that the winners won’t be firms that simply add AI features to existing processes, but those that fundamentally reimagine how core functions like underwriting, development approvals, or asset management work from start to finish.
