The real estate sector is complex and produces reams of information including property, transaction, and market data. Nevertheless, technological superiority has not traditionally been an important competitive advantage for participants. That likely explains why AI adoption, which has been rapid in sectors like technology and finance, has proven slower in real estate--though we expect the pace will soon accelerate.

Real estate companies' limited use of AI has, until recently, focused on tenant management systems, asset valuation, and price forecasting. Elsewhere, long-standing practices, including complex operating processes and human interactions, have often proven resistant to newer technological solutions. And the local and fragmented markets in which real estate companies generally operate have inhibited the deployment of technological solutions that are universal and scalable.

Yet recognition that AI could add value across the industry is growing. S&P Global Ratings believes that the benefits (and risks) that are inherent to AI's development and application will increasingly affect real estate companies' operating efficiency, competitiveness, and credit quality.

We see myriad applications for AI across the real estate sector's diverse spectrum of operations. For example:

• Discriminative AI (which applies statistical algorithms to perform tasks, see "Machine Learning: The Fundamentals," Nov. 29, 2023) could expand its role in market analysis to produce increasingly customized and dynamic pricing and improved forecasting. Machine learning could also find new applications in construction site management, notably to minimize waste, speed operations, and reduce costs.
• Generative AI (which uses algorithms to interpret data and produce new outputs, see "Foundation Models Powering Generative AI: The Fundamentals," Nov. 29, 2023) could enhance efficiency by improving the timeliness and pertinence of reporting, identify investment opportunities, improve building design, and reduce costs by automating marketing and sales functions.

Over the next five to 10 years, we expect AI will be an increasingly important part of the foundation on which real estate companies' credit quality is built. For that reason, we will closely monitor how the technology is applied and the outcomes it delivers in the real estate sector.

AI Adoption Faces Real Estate Sector-Specific Hurdles

The real estate sector's relatively slow embrace of AI is the result of characteristics that affect various sections of the industry differently, but which have combined to generally inhibit adoption (see table 1).

Table 1

Uses For AI In Real Estate Will Be Numerous

While the real estate sector has some characteristics that have inhibited the adoption of AI (see table 1), there are also many tasks and problems that are suited to the application of AI and AI-enhanced solutions (see table 2).

Table 2

AI's Impact On Real Estate Asset Classes Will Vary

Not all subsectors of real estate will use AI in the same way, to the same extent, or realize the same benefits from the technology. Most evidently that is due to the significant differences between companies and across the real estate subsectors in which they operate. For example, market characteristics, both in terms of geography and asset type, can determine if data sets are sufficiently large and accurate, with implications for AI model training and thus the veracity of output. And safety requirements, which remain paramount across the sector, ensure that AI's application is corralled by both regulation and best practice, both of which will differ from region to region. The diversity of AI adoption across real estate sub sectors will reflect competitive pressures, capital expenditure capacities, ingrained development and property management practices, and the characteristics of various types of assets (buildings and portfolios).

Data centers

We expect near-term demand for data centers will be fueled by the need to house the hardware on which AI systems operate. Generative AI, in particular, is a strong demand driver for data centers and growth in both model training and inference (where a trained machine-learning model draws conclusions based on data) have proven to be significant contributors to data centers' revenues. We expect demand for hyperscale data centers will accelerate in line with the growing requirement for large-scale and high-powered platforms to power AI computing. Limited supply of data centers, due to energy bottlenecks, high construction costs, and planning and regulatory impediments, is affecting the supply/demand balance. We observe that large well-capitalized tech companies are making significant investments in proprietary data centers, but we also expect sustained demand for rented capacity from data-center operators.

Industrial

Logistics and warehousing will likely be reshaped by increased automation, which will both reduce employment in the segment and alleviate much of the need to locate facilities near manual labor. AI solutions, including imaging and vision technologies in combination with predictive analysis, will improve inventory planning and optimize space usage and could thus alter space requirements at existing facilities. This may create a polarization between technologically-advanced facilities and more obsolete ones.

Residential

The significant amount of data and relative uniformity of many assets means the residential sector is suited to the application of machine learning (including in predictive analytics and price discovery). Smart buildings systems, which are increasingly common in residential houses, are likely to benefit from AI development. AI could also be applied to improve developers' housing stock provision, notably through the AI-enhanced customization and matching of housing development projects to factors including demand, supply, and social requirements.

Use of AI in the mortgage industry is already increasing and could lead to significant changes for both lenders and borrowers. That could include material cost savings from automation, improved performance due to AI-enhanced decision making, highly-personalized mortgages, and faster processing (see "AI In Banking: AI Will Be An Incremental Game Changer," Oct. 31, 2023).

Retail

AI will improve the experience of virtual shopping, possibly at the continued expense (and in some instances the demise) of physical outlets. Generative AI will accelerate this shift with new services, including virtual fitting rooms and tailored recommendations. Retail real estate is likely to seek to compete by prioritizing the quality of its experience, for example through the creation of destination shopping centers. At the same time, AI can help retail landlords better anticipate individuals' behavior, customize marketing, and improve the shopping experience to capitalize on changing consumer habits.

Office

AI could prove a catalyst for increased remote working, reinvention of jobs, and workforce reskilling, all of which may reduce or alter demand for existing workspaces. Remote working has already resulted in changes in occupancy, office utilization, and the size of leases. Generative AI's ability to facilitate remote but collaborative working is likely to exacerbate that trend (see "Metaverse and Generative AI: Envisioning the Future of Human-Computer Interaction," Nov. 7, 2023). At the same time, it may also help create new jobs. Generative AI could be applied to find novel ways to convert underused office space into new assets (see following section: Generative AI Will Make Inroads).

Generative AI Will Make Inroads

The real estate sector's characteristics have inhibited the use of some forms of AI, and encouraged others. For example, AI adoption by the sector has favored discriminative machine learning and deep learning models. That is because discriminative algorithms offer higher predictive accuracy, and are also generally simpler, more interpretable, and more scalable (compared to generative models), which makes their implementation more cost-efficient. Discriminative AI is also propelled by the richness of data in the industry, clear boundaries between asset subsegments and price ranges, and the need for unambiguous and highly-interpretable modelling outcomes.

Deployment of AI as a value-added tool for data entry and number-crunching will, in our view, likely remain a prominent feature of the real estate sector. AI's growing capabilities coupled with larger, and often bespoke, data sets should deliver new uses and better insights from those models. Yet the sector is also increasingly open to more creative possibilities arising from generative AI and we expect to see that technology increasingly deployed to meet specific challenges (see chart 1).

Chart 1

AI Opportunities In Real Estate We Are Watching

We believe the real estate sector can significantly benefit from the implementation of AI, and expect those opportunities to grow with development of the technology and as confidence in its promise increases. Many of the applications of AI in the sector will be general in nature, and mirror their use outside of the real estate sector, though no less powerful as a result. For example, AI-enhanced image creation and three-dimensional modelling, both sure to be widely adopted across many sectors, will enable virtual- and augmented-reality tours of finished properties, planned developments, work sites, and sites for potential development.

Yet there are also technological opportunities involving AI that are specific to the real estate sector (see "The Rise of AI-Powered Smart Cities," May 18, 2024). Some of these have the potential to significantly affect the industry by establishing competitive advantages for early and effective adopters. Among the developments we are particularly watching are:

Property management technology (proptech),  which promises to streamline and reduce costs associated with the management of multiple real estate elements, including tenant portfolios, facilities, finance, and risk. The multifaceted nature of proptech means it will likely be underpinned by composite-AI technology, including elements of generative AI (that could drive chatbots, virtual assistants, and reduce payroll costs) and predictive analytics (which might be applied to forecasting or reducing vacancies by matching rentable spaces with potential tenants). Proptech is already being used by, and will be further developed for, companies operating in real estate markets. For example, AI-driven dynamic pricing models are increasingly applied to house pricing and for setting rents in the flex-office and self-storage markets.

Building information modelling (BIM),  which is likely to increasingly integrate AI into design tasks ranging from generating office layouts, creating building floor plans, and rendering three-dimensional models of planned constructions for virtual tours/staging. Generative AI will play a lead role in facilitating the interface between humans and AI systems.

AI-based solutions for urban planning and smart cities,   which could support decision-making for real estate development. AI-based applications for smart cities include traffic system management, crime detection, air quality monitoring, and environment-friendly solutions. Additionally, AI can be used to improve the efficiency and scalability of building and site plans by adjusting them to differing planning, zoning, and code regulations, particularly where projects span areas subject to different governing bodies.

AI-enhanced assessment of remote area, brownfield, and inefficient/stranded assets,  should offer real estate developers improved data collection, and analysis, including of local (down to neighborhood-level) information that can be used to guide new development, revitalization, and retrofitting decisions. Such assessments can also determine factors such as optimal mixes of affordable versus market-rate housing supply and the provision of mixed-use areas that incorporate residential, retail, and commercial areas. Much of the data assessment will be conducted by discriminative machine learning and deep learning models, while Generative AI will likely find a role in reporting and stakeholder communications as well as providing new solutions for risk analysis and management (including climate risks scenarios) and asset quality improvement.

Optimization of preventative maintenance and capital expenditure  using AI-driven models, which can offer owners the chance to increase an asset's lifespan, reduce costs through more timely and effective intervention and preventative maintenance, enhance an asset's appeal to tenants or would-be buyers, and improve buildings' energy efficiency. Similarly, AI could be used to predict emerging/best use cases for underutilized/financially underperforming assets. That could include, for example, the conversion of vacated or struggling commercial areas into residential areas or assets with high social value (such as schools, shelters, healthcare facilities, or affordable housing)--with the viability assessment, solutions, and resulting plans all created by generative AI.

Management through digital twins,  which offers property developers and managers the opportunity to experiment with real-life properties by creating virtual doppelgangers, either in the form of a visual representation or a dataset. Because twins are created from huge amounts of real-time data, including from sensors and cameras, costs may be substantial. But the benefits could be material, and include the opportunity to virtually and rapidly test alternative space configurations and new systems (including heating, ventilation, air conditioning (HVAC), and security). Twins can also be used to stress existing assets to predict failures, mitigate the impact of weather events, and improve property management and customer service.

AI And The Future Of Real Estate Sector Creditworthiness

The relatively nascent nature of AI's application to real estate, and the near-term difficulties in applying AI solutions, which we detailed earlier, means that the financial benefit of an investment in AI by real estate sector actors is likely to be gradual. We expect that our credit analysis will increasingly integrate AI factors as the effects of the technologies' adoption become apparent.

Issuers with clearly defined and articulated AI strategies, governance frameworks adapted to the their business strategy, and which demonstrate successful integration of AI in their operations stand to reap the greatest benefit. Furthermore, AI is likely to serve as a force-multiplier, providing those that secure a competitive advantage the opportunity to accelerate away from rivals in terms of revenues and margins, better forecasting, and improved asset quality.

While AI adoption is expected to create opportunities for real estate companies, it also comes with risks, which, if crystallized and material, might impact our view of companies' creditworthiness (see table 3).

Table 3

Related Research

S&P Global Ratings research

• The Rise Of AI-Powered Smart Cities, May 18, 2024
• Language Modeling: The Fundamentals, Jan. 23, 2024
• Can Generative AI Create A Productivity Boom?, Jan. 10, 2024.
• Machine Learning: The Fundamentals, Nov. 29, 2023
• Foundation Models Powering Generative AI: The Fundamentals, Nov. 29, 2023
• Metaverse and Generative AI: Envisioning the Future of Human-Computer Interaction, Nov. 7, 2023

Other research

• Generative Adversarial Networks In Construction Applications: Automation In Construction (159), Chai, P., et al., 2024.
• Generative AI In The Construction Industry: Opportunities & Challenges, Ghimire, P., et al., 2023.
• An Image Is Worth 16x16 Words: Transformers For Image Recognition At Scale, Dosovitskiy, A., et al., 2021.
• Complete Guide To Digital Twin Technology in Construction, Nishanth, PK., 2023.

Writer: Paul Whitfield