AI's Impact on Office Demand

This is the second in a four-part series of reports by CBRE Research on the impact of artificial intelligence on commercial real estate. This report examines AI’s impact on the office sector, including its influence on office-using job growth and which markets are best positioned to benefit from AI-driven demand.

You can read Part 1 here: AI’s Impact on the Economy, Employment & Productivity

Despite concerns about widespread job losses from artificial intelligence utilization, the underlying data indicates otherwise. Private sector layoffs are near their lowest rate since 2013. It's hiring that has been affected most: Job openings have declined sharply from their 2022 peak, creating a low-hire, low-fire environment where workforce churn has broadly slowed. Even as the private sector moves past this period of economic- and AI-driven uncertainty, the real culprit behind slowing office job growth is demographics. An aging workforce resulting in roughly 83,000 new retirees per month, combined with sharply curtailed immigration, is the more powerful force constraining job growth in the long-run. High-profile tech layoffs, meanwhile, largely reflect companies unwinding pandemic-era overhiring rather than an AI-driven reckoning.

Figure 1: Job Openings, Hires & Separations

With both hiring activity and labor supply constrained simultaneously, a slower pace of office job growth is the natural consequence. The result: Office-using employment is forecast to grow at just 0.9% annually over the next five years compared with a 1.5% long-run average and mainly driven by labor supply constraints, not AI displacement. The silver lining is that this 0.9% projection is stronger than the forecast 0.6% job growth rate for the broader economy, reinforcing that office-using industries remain among the biggest job creators.

Figure 2: Compounded Job Growth

AI is both changing existing jobs and creating new ones. Here are three reasons why this is resulting in more opportunity than disruption.

01. Those Most at Risk Are Also Most Adaptable

As noted in Part 1 of our AI Impact series, recent research from Manning & Aguirre (NBER, 2026) finds a strong correlation between AI displacement risk and adaptive capacity. The white-collar workers most exposed to AI disruption are also the best equipped to harness the productivity gains of new technology. Job disruption and creation are not mutually exclusive— historically, they have gone hand in hand.

The Jevons Paradox explains why: As AI makes services cheaper and faster to deliver, aggregate demand for those services rises, thereby generating more jobs. Consider an analyst vetting investment opportunities. As AI reduces the time-per-deal review, the firm can perform due-diligence on more deals in less time, increasing its number of business opportunities.

The result is more demand for the service offering. A surge in weekly business formations in recent years is evidence of this, driven in part by AI making it cheaper and easier than ever to start a company.

Figure 3: New Business Applications

02. History Shows Technology Creates Office Jobs

Following the advent of both the internet and the smartphone—technologies that generated remarkably similar fears of white-collar displacement—office-using jobs had their largest share of overall employment growth since at least 1990. The reason is structural: Work created by new technology is more productive than the work it displaces. Office-using industries generate output per employee more than twice the national average, and that productivity premium has been widening even as broader productivity growth has stalled. As AI drives the next productivity boom, it will do what prior technological shifts always did: push labor toward the most productive sectors of the economy, which skew heavily office-using.

Figure 4: Overall vs. Office-Using Employment by Decade

03. The Jobs AI Eliminates and the Jobs It Creates Both Highlight the Importance of the Office

The roles most vulnerable to AI automation are disproportionately those that already are mostly remote. These jobs share two traits: They don't require in-person collaboration and their performance can be measured through well-defined tasks and clear benchmarks. Those are also the characteristics that make work automatable.

As AI absorbs routine task-based functions, it frees capacity for something harder to automate: complex, judgment-intensive, collaborative work that is best performed in person. The implication for office demand is straightforward: The work being automated skews toward routine functions that have largely left the office already. The work being created favors roles that need to have a physical presence, pointing to a net tailwind for office demand by the most collaborative roles.

Figure 5: AI Exposure vs. Remote Work by Occupation

While the macro picture is becoming clearer, two important questions remain: How will AI's impact vary across industries? And which markets and property types will benefit most?

Manning & Aguirre's research assigns every U.S. occupation two scores: an AI Automation Risk score that determines how likely that role is to be disrupted and an AI Adaptive Capacity score that shows how well that role can benefit from AI-induced change. Mapping these scores across traditional office-using sectors reveals which industries are more likely to expand versus consolidate as AI adoption accelerates.

The main finding is that just 5% of the office-using workforce is considered highly vulnerable to AI displacement, while 18% is characterized as highly adaptive. The remaining 77% sits somewhere in the middle—exposed to change but not necessarily displacement.

Figure 6: Breakdown of AI Vulnerability & Adaptability by Office-Using Industry

The information sector, which accounts for much of the country's technology workforce, has the lowest share of vulnerable workers and the highest share of highly adaptive workers of any office-using segment. The professional, scientific & technical services sector also has a large share of tech and knowledge workers, helping to explain its highly adaptive nature.

The finance, professional services and administrative services sectors have the greatest number of employees vulnerable to AI displacement, although just 6% of their collective workforce is exposed. Within each of these sectors, there is a wide distribution of skillsets and therefore a wide distribution of outcomes.

The job makeup of industries can look very different from market to market. The mix of finance jobs in New York bears little resemblance to the mix in a mid-sized back-office market. Mapping occupational breakdowns at the metro level can identify where the same industry's AI vulnerability is highest and lowest. This has direct implications for occupiers assessing where their workforce is best positioned and for investors evaluating the resilience of their tenant mix.

Take financial services, which has a relatively high share of vulnerable workers. On its face, this might suggest that major financial hubs are most exposed to a downturn in office demand. In reality, the opposite is true: The most vulnerable financial services roles are concentrated in back-office markets. The metros with the most resilient finance sectors are New York, San Jose, Charlotte, San Francisco, Boston and Salt Lake City.

A similar pattern is apparent for professional & technical services, another sector with elevated vulnerability nationally. The major hubs of San Francisco, Washington, D.C., Austin, Boston and Denver have the most resilient workforces within that segment.

Looking at total office-using employment, two Bay Area markets—San Jose and San Francisco—rank as the least vulnerable metros to AI automation, followed by Washington, D.C., Seattle and Boston. These markets share certain characteristics: a highly educated workforce with adaptable skillsets that are best positioned to harness AI as a growth driver.

Figure 7: Least Vulnerable MSA Markets by % of Occupied Office Space at Risk by AI Disruption

Figure 8: Relative AI-vulnerability of the Office Workforce by MSA

AI's impact on where work happens will be a bigger driver of office demand than on the number of jobs—a more positive outcome than conventional wisdom suggests.

As AI absorbs task-based, process-driven work, the balance of office-using roles will shift toward the more nuanced, collaborative work that is best performed in person. Fully remote roles, which make up approximately 13% of the U.S. workforce, are inherently more vulnerable to automation.

A recent Live Data Technologies analysis of 2 million white-collar workers found that fully remote employees were 35% more likely to be laid off in 2023 than their hybrid or in-office counterparts. This suggests the prospect of historically mild office-using employment growth and a rising share of workers that perform their work in-office, which in turn will increase the space-per-worker ratio. This is already happening: Since the mass introduction of AI tools in 2023, office-using employment has been flat but space-per-worker has been on the rise.

Figure 9: U.S. Occupied Office Space Per Worker

AI is also lowering the barrier to business formation, driving increased demand from smaller users and flexible space providers. Both trends reinforce the same conclusion: Dense centers of business are best positioned to benefit. In markets where more workers are well-positioned to benefit from AI adoption, office occupiers would be wise to avoid over-consolidation— particularly as demand for prime space increasingly outstrips available supply.

CBRE's proprietary database of lease transactions reinforces this conclusion. Technology companies accounted for 21% of total U.S. office leasing in H1 2026, on par with their highest share on record. Nearly two-thirds of that activity was in the San Francisco Bay Area and Manhattan, the most concentrated distribution on record and reiterating the attraction of major business hubs.

Lease terms among tech users have also lengthened substantially, particularly among the largest occupiers. All of this points to one clear conclusion: Those companies that best understand the potential benefits of AI and stand to be most transformed by it are increasing their real estate footprints.

Figure 10: Average Office Lease Length for Tech Occupiers

As labor continues to shift toward more productive roles, the highest-quality office assets that are increasingly in demand for these highly skilled workers will benefit most. With new construction pipelines dramatically reduced, the U.S. office market is on track to effectively run out of prime space by the end of 2027. Our baseline forecast reflects this dynamic: The overall office vacancy rate is expected to fall to 14.5% from 18.3% by 2031, led by urban hubs. Drivers of this performance include improving job growth, more demand for a dwindling supply of quality space and stronger occupier sentiment.

Given today's uncertainty, we've also constructed a theoretical downside scenario based exclusively on AI's negative potential impact on jobs. Just 5% of the office-using workforce is considered highly vulnerable to AI-induced displacement. Under pessimistic assumptions that all of these workers are today contributing to office demand and that all of them will be displaced, the net impact would add just 0.4 points to the current overall vacancy rate, bringing it to 18.7% by 2031.

Even in this downside scenario, markets with the most resilient workforces—the San Francisco Bay Area, Washington, D.C., Seattle, Boston and Denver—would be most insulated from AI disruption. Prime office space also would continue to exhibit the strength we're seeing today, as workers that occupy such space are among the least vulnerable to AI displacement.

Figure 11: Office Vacancy Rate with Theoretical AI Downside

As AI increases productivity, economic theory and history suggest it will be accretive to office-using job growth, mirroring the impact of prior technological advances.

Jobs that will be eliminated by AI are largely remotely performed, task-based roles, while those that will be created by AI involve largely collaborative in-person work. That substitution could prove a meaningful tailwind for the office sector.

Location and industry mix will determine which markets will benefit most. Metros with highly skilled, educated workforces are best positioned and industry differences among markets matter greatly. For example, despite financial services' relatively high vulnerability nationally, New York has the least exposed financial sector of any major U.S. metro. Occupiers and investors alike must resist broad-brush conclusions and engage at the market and industry level.

Ultimately, AI will accelerate the office market's K-shaped recovery: rent growth for the prime office market that serves the most productive workers vs. hastened obsolescence for secondary space as the value of simply housing workers continues to erode.