AI and the Future of Office: Quantifying Workforce Change and Space Demand Through 2030

Part I: Labor Trends and AI’s Workforce Impact

AI, Fear and Familiar Patterns of Disruption

Intense hype around AI‑driven productivity gains has been accompanied by increasing concerns about job displacement, echoing anxieties from past technological transitions. Ultimately, the outcomes will depend on several factors, including the pace of AI adoption, the balance between task augmentation and automation for various occupations, how firms choose to apply productivity gains and how demand for goods and services evolves.

Pace of Adoption  

AI’s impact on jobs will depend on how quickly and how deeply organizations move from AI pilots to fully redesigned workflows. Today, the barrier to realizing AI’s potential lies not just in the technology itself, which is improving at a rapid pace, but in the slow pace of scaled adoption, due to factors including distrust, regulatory complexity and unclear governance and risk standards. McKinsey’s 2025 global AI survey finds that while 88% of organizations use AI in at least one business function, 62% remain in experimentation or pilot mode, and only about one‑third have begun to scale AI across the enterprise, primarily larger firms with more than $5 billion in revenue. Similarly, a recent Anthropic report comparing theoretical task exposure with observed usage on its Claude platform finds a wide gap between theoretical capability and adoption. In computer and math occupations, for example, large language models could theoretically perform 94% of job tasks in that category, yet current usage covers only about 33%. Despite this early stage, business leaders are betting on fast and far-reaching change. The 2025 World Economic Forum (WEF) Future of Jobs report reveals that 86% of employers expect AI and information-processing technologies to transform their business by 2030, suggesting today’s pilots are laying the foundation for large‑scale change.

Augmentation vs. Automation 

AI will affect jobs through a mix of automation and augmentation. Automation occurs when AI fully takes over tasks previously performed by workers, with limited human oversight. Augmentation, on the other hand, involves AI complementing and enhancing a human’s work, allowing workers to focus on their areas of comparative advantage, thereby increasing the quality and volume of output through greater specialization. Occupations with a higher degree of automatable tasks have a greater displacement risk than jobs that are more highly augmented.

Based on survey responses from over 1,000 global employers, the WEF report estimates that in 2025, an average of 47% of work tasks across occupations were performed solely by humans, 22% by technology and 30% by a combination of both. By 2030, employers expect these shares to be nearly evenly split (Figure 3). However, these values vary substantially by sector and occupation.

Evidence suggests that current AI use leans more towards augmented rather than automated work. Anthropic’s January 2026 Economic Index found that 52% of Claude conversations were classified as augmentation compared with 45% categorized as automation, with augmentation especially prevalent in complex, knowledge‑intensive tasks. If this pattern holds as adoption deepens, many more jobs are likely to be restructured through augmented workflows than replaced by automation, which would temper but not eliminate displacement. Achieving this outcome will require sustained public‑ and private-sector investment in upskilling workers and training a new pipeline of talent to work effectively alongside AI. 

Reallocating AI Efficiency Gains 

What companies, and society, do with AI‑driven efficiency gains will be critical for labor‑market outcomes. For our purposes, these gains come in two forms: time and money. Historical experience shows that redeploying human capital and the real income generated by positive productivity shocks can be powerful. Each major wave of new production technology—steam power, electrification, mechanization, computing and the internet—ultimately gave rise to new industries and occupations. Despite near‑term disruption, these advances lifted overall employment and standards of living over time. As Citadel Securities observed in a recent report, “A scenario in which productivity surges but aggregate demand collapses while measured output rises, violates accounting identities.” 

A 2024 McKinsey survey finds that most organizations reallocate time saved by automation into new activities or higher‑value work. Larger employers, however, are more likely to reduce headcounts as efficiencies build—and those reductions are strongly associated with greater bottom-line gains from generative AI. The balance between reinvestment and role redesign versus layoffs will likely be an important factor in shaping net employment outcomes. 

These efficiency gains will likely spur new firm creation, which has already surged over the past few years. Annual business applications in the U.S. are well above pre‑pandemic norms and more than double the 2018–2019 average. By lowering the cost, time and expertise required to start and run a business, AI can further accelerate new business creation. 

Office Sector Exposure to AI

AI’s impact on employment will be uneven across industries and occupations. Knowledge‑based roles—especially repetitive or entry‑level positions typically performed in offices or remotely, are among the most exposed, as AI increasingly conducts basic administrative, analytical and communication tasks with little human oversight. Traditional “office‑using” sectors— professional and business services, information and financial services—are heavily composed of these knowledge roles (telemarketers, insurance claim and policy processing clerks, credit authorizers, payroll staff, travel agents, tax preparers, bill collectors). In contrast, service and field‑based occupations that take place primarily outside the office (athletes, paramedics, landscapers, roofers, lab technicians, barbers) are expected to be less affected given their emphasis on in‑person, manual or context‑rich work that is harder to fully automate.

Figure 5 illustrates select industries and their expected shift in the human share of work‑task delivery in total firm output between 2025 and 2030, distinguishing between changes driven by automation and those driven by augmentation, based on WEF survey data. The steepest declines in human‑only task share are concentrated in office‑based sectors. In professional and business services, for example, employers estimate that the human-only task share will drop from roughly one‑half today to about one‑third by 2030, with approximately two‑thirds of that decline attributable to outright automation rather than augmentation. We incorporate this sector‑level exposure to automation and augmentation into our broader model of AI’s impact on employment and office demand, discussed later in this paper.

Because AI exposure is highest among office-using occupations, and office demand depends on the growth of those jobs, the office sector faces the greatest exposure risk among commercial real estate property types. As AI adoption accelerates, slower job growth in office-using industries could further weigh down future demand for office space. In 2024, Oxford Economics reached the same conclusion, as seen in Figure 6.  

Current Evidence of AI-Related Job Displacement

AI has become a defining theme in how companies describe and strategize around their workforces, yet its effects are hard to discern in labor market data. To gauge what’s happening now, we first examine how employers are framing AI’s role in job design, skills and workflows, and then compare those narratives with the underlying employment data.

What Employers Are Saying 

Several prominent employers—including Amazon, Walmart, Salesforce, Block, and CrowdStrike—have referenced AI-enabled efficiency gains in public statements related to recent layoffs. Analysts caution against interpreting these disclosures as evidence of AI-driven job displacement, noting that workforce reductions are often influenced by broader factors such as overhiring, slowing demand or margin pressures. To the broader public, however, this framing may lead many to overestimate AI’s immediate labor‑market impact.

The Federal Reserve’s January 2026 Beige Book tempers the narrative of an AI-driven hiring collapse. Multiple contacts across districts reported exploring AI primarily for productivity enhancement and potential future workforce management. They noted incremental productivity gains to date and limited near-term effects on headcount, with more meaningful changes expected over the coming years. Marketing, call center and coding roles were cited as at‑risk, although often in offshore locations, and some firms indicated that modest AI adoption allows them to avoid refilling positions, leaving vacancies unfilled through natural attrition.

Leaders of AI firms have publicly forecast significant white-collar workforce transformation, sentiments that seem quite sober compared to some of the more outlandish prognostications floating around in the discourse. While these perspectives highlight AI’s potential, they also reflect narratives intended to position AI as an essential investment that reduces labor costs. Given this context, it is important to interpret these types of forward‑looking statements alongside empirical labor‑market data and observed adoption trends, recognizing the substantial uncertainty surrounding AI’s long‑term effects on the workforce.

What Labor Data Shows 

Next, we need to turn to the labor market data to gauge the impact of AI on jobs. The U.S. labor market has seen relatively weak growth over the past two years, with a low-hire, low-fire dynamic and office‑using employment being essentially flat in 2025 (Figure 7). While the economy has not experienced large-scale layoffs in recent years, companies still trimmed headcount and slowed or paused hiring, and fewer employees chose to leave their jobs voluntarily.

Most net job growth in 2025 came from non‑office‑using, lower‑AI‑exposure sectors—led by education and healthcare, which now employs about 13% more workers than before the pandemic. By contrast, the information sector—an office‑using industry with higher AI exposure that includes many technology workers—still has fewer jobs than in February 2020. This labor market slowdown, outside of healthcare, began before the recent wave of AI adoption and largely reflects structural supply‑side constraints, such as an aging population, tighter immigration and the information sector’s pandemic‑era over‑hiring and subsequent layoffs.

Current data suggests that AI is exerting a modest and diffuse effect on the overall labor market. It appears to be a contributing factor rather than the main driver of recent softness, and its impact remains difficult to disentangle from other underlying forces—an important caveat to keep in mind throughout this discussion. 

Early signs of AI’s labor impact have appeared among younger workers in highly AI-exposed roles. A 2025 study by the Stanford Digital Economy Lab found that employment declines were concentrated among 22–25‑year‑olds in exposed occupations such as software development, customer service and clerical work, while employment for older workers in the same occupations, and for workers of all ages in less‑exposed roles has remained stable or continued to grow. 

These patterns help explain why overall employment for 22–25‑year‑olds has been relatively flat since late 2022, even as the broader labor market has expanded. In less AI‑exposed jobs, young workers have kept pace with their older counterparts, but in highly AI‑exposed occupations their employment fell 6% from late 2022 to September 2025, while older workers in the same occupations saw gains of 6-9% (see Figure 8). 

Impact on Real Estate: The Office Space Equation

The Last Most Recent Structural Transformation: Hybrid Work Effects 

The AI revolution is unfolding against the backdrop of an office market that has only recently regained its footing after the disruptive shift to hybrid work. That transition drove 18 consecutive quarters of negative net absorption and pushed the U.S. office vacancy rate to a record 20.6% in Q2 2025. Just five years ago, hybrid and remote work were viewed as existential threats to the office sector, a “sky-is-falling” moment that many feared would spell the end of the workplace as we knew it. In hindsight, hybrid work reshaped the office rather than erased it.

Today, the market is distinctly bifurcated: trophy and Class A space continue to outperform, while commodity space lags well behind. Newmark maintains a database of tenants with open space requirements in a wide range of national office markets. Presently, 70% of active requirements involve maintaining or expanding the tenant’s existing footprint upon lease expiration. Flex Index data reinforces this resilience: roughly 95% of companies still operate at least one office location, while only about 5% are fully remote (Figure 10). Similarly, Indeed data show that job postings mentioning remote or hybrid options remain a modest but durable share of all listings at 8.9% as of February 2026, down slightly from a pandemic‑era peak of 10.3% in early 2022.

Hybrid work has nonetheless permanently altered office-usage patterns. Attendance has stabilized at about three days per week, and the office has evolved from being the default worksite to a hub for collaboration, connection and team building. This shift has trimmed space needs, with occupied square footage per office worker falling about 10% between early 2020 and 2025. Yet market fundamentals have begun to stabilize. Move-ins once again outpaced move-outs in 2025, and a slowdown in construction is paving the way toward a more balanced market.

Early-pandemic era fears of an “office collapse” which echoed the “death of retail” narrative that accompanied the rise of ecommerce over a decade ago, were ultimately overstated in both cases. Likewise, current conversations around AI and automation risk seem to be mirroring that pattern.

AI Is a Positive Trend for Near-term Office Demand in Select Markets 

Artificial intelligence is currently a major catalyst for office demand, particularly in the San Francisco Bay Area, the capital of tech talent and venture investment. In the near term, this momentum is poised to continue as rapid AI adoption and sustained capital inflows spur new company formation and headcount expansion. AI firms now account for a substantial share of San Francisco’s net absorption, helping reduce office vacancy for the first time in years. Collectively, these firms occupy more than 7.5 million square feet of space. Another 4,500 AI startups are headquartered in San Francisco but operate out of apartments or flexible workspaces, representing a sizable pipeline of future demand as they scale. Active AI space requirements exceed 2.6 million square feet and include expansions from firms such as OpenAI, Harvey, Together AI, Perplexity, Sigma, Cursor AI and Cognition AI, while Sierra and Anthropic have collectively leased more than 700,000 square feet in recent months.

Momentum is also growing in Manhattan, where AI and tech-driven demand are becoming increasingly central to leasing dynamics. Tech and media tenant requirements climbed to their highest level since 4Q18 at approximately 8.5 million square feet, with AI-related demand representing 10-11% of that total—more than double the sector’s share just two years ago.

Part II: Quantifying the Impact—Our Modeling Approach

Employment and Office Market Outcomes Under AI Scenarios

To estimate AI’s impact on office-using jobs, we used a Monte Carlo simulation based on a model that translates AI adoption and task automation into potential labor -market impact (see Appendix for full model details). The analysis draws on data from the WEF Future of Jobs 2025 report, which provides estimates of the share of tasks performed exclusively by humans, by technology, and the extent to which AI automation is expected to reduce human-performed tasks by 2030.

We conducted the analysis at the sector-level rather than the occupational-level, to better reflect how differences in AI investment, digital readiness and regulatory environments affect the timing of potential job displacement by 2030. Technology firms, for example, are likely to adopt AI faster and reshape roles sooner than slower moving sectors like oil and mining, even where similar tech-oriented occupations exist in both. Using sector-level analysis therefore provides a clearer view of when AI-driven changes to jobs are most likely to occur over the short forecast horizon, albeit at the cost of some occupational granularity. Each of the three major office-using sectors was assigned an automation-potential value from the WEF report, representing the share of tasks that could be automated by 2030. We then estimated the relationship between automatable tasks and job loss. For instance, if 50% of tasks are automated, does that translate to a 50% reduction in jobs (ratio = 1) or a 25% job reduction (ratio = .5). We also estimated the positive employment effects that come from AI-driven productivity gains, using McKinsey’s estimates of potential output growth. Based on World Bank data, we assumed that each 1% increase in output translates into a 0.45% increase in employment.

These inputs were run through 1,000 simulations over a five-year period, allowing the model to vary key inputs including AI productivity boost, ratio of task to job losses, adoption rate of AI and regulatory friction, which produced a range of possible outcomes. We largely focused on the variation in the ratio of task to job loss, as that tends to drive the results. For each scenario, office-using employment was converted into space demand using the current average square footage per employee. Comparing these demand estimates to the expected office-supply pipeline allowed us to calculate U.S. office vacancy rates in 2030 under different AI adoption paths. Further detail on the methodology can be found in the Appendix.

We present five distinct scenarios below, including a baseline case, to show how office employment and space demand could evolve through 2030.

In the control scenario, which does not explicitly incorporate AI’s impact on employment, Oxford Economics projects office-using jobs will grow by 2.4% between 2025 and 2030. In contrast, Newmark’s Base Case AI scenario forecasts employment growth will be essentially flat (+0.3%) over the same period. While this outcome is far from the “AI job apocalypse” portrayed in some headlines and online discussion boards, it does represent a break from historical trends. While office-using employment has experienced flat or declining periods before, most notably during the Great Recession, this has never occurred, in data going back to 1944, without an accompanying recession. However, stagnant growth over a similar timeframe is not unprecedented in other employment sectors. Manufacturing employment in the United States peaked in 1979 and fell precipitously between 2000 and 2010. In several periods, manufacturing employment fell without a recession, including the five years ended December 2007, March 2001 and July 1990. Meanwhile, industrial production by the manufacturing sector has increased 108% since 1979 as productivity transformations enabled greater output without increases in labor. Ultimately, technological shifts simultaneously drove the emergence of new industries, jobs and facilities within manufacturing—expanding the sector’s overall real estate and demand footprint even as its labor composition evolved. The economy also adjusted by creating new jobs and shifting labor resources to other sectors, namely office-using sectors and the knowledge economy. A similar dynamic could define the trajectory of office-using employment and industries in the age of AI. 

Office Market Impact 

Building on the five scenarios outlined for our office‑using employment forecast, we next translate those labor market outcomes into projections for U.S. office vacancy. We find that the U.S. office vacancy rate in 2030 varies depending on how artificial intelligence reshapes office work and space demand, with a roughly 4 percentage point difference between our most optimistic (moderate upside) and pessimistic (severe downside) scenarios (Figure 12). In the moderate upside scenario, vacancy edges down by about 30 basis points relative to our control scenario with no explicit AI effect, falling to 19.5%. At the opposite end of the spectrum, in the severe downside scenario, vacancy rises by as much as 360 basis points above the control, reaching 23.5% by 2030, a record-high for the U.S. office market. Our base case sits somewhere in between, with vacancy increasing 170 basis points relative to the control, or 10 basis points above the Q4 2025 level, at 21.5%. 

To estimate vacancy, we examined the relationship between office-using employment and occupied space. As of 2025, traditional office-using jobs in the United States supported 192 square feet of occupied space per worker—a ratio that has held relatively steady in recent quarters and that we interpret as the post‑pandemic equilibrium following the normalization of hybrid work. We applied this ratio to projected office‑using employment levels in each scenario to calculate total occupied space in 2030. Combining these figures with our outlook for total office inventory, based on known construction pipelines and delivery schedules, enabled us to derive the implied vacancy rates under different AI adoption trajectories. This step‑by‑step approach links labor‑market outcomes directly to office‑market fundamentals, illustrating how the variation in these scenarios could reshape the office market by the end of the decade. 

Takeaways for Commercial Real Estate Stakeholders

For Investors

1. Quality Office Will Become More Desirable

Post-pandemic, demand for office has concentrated on the highest-quality office space. Midtown Manhattan trophy assets, for example, have a direct availability rate of 3.7%, compared with a market-wide rate of 15%. As office-using jobs become more productive, and competition for talent intensifies, quality office space will be a critical tool for attracting the best workers. In an environment where total office demand may decline, whether due to reduced headcount or lower space per employer, location and quality become not only drivers of returns but key differentiators for outperformance.

2. Agglomeration Economies Will Matter Even More

Markets that attract deep pools of talent, such as Silicon Valley for tech and New York City for finance, tend to outperform in periods of stagnant demand growth. These agglomeration economies, or clusters of industries that tend to drive innovation, act as fortress markets in the long run, as market performance expands and contracts from these epicenters, and often comprise the CBDs of their markets. Assets near the epicenter have better demand drivers and tend to recover earlier as credit tenants are more concentrated. Agglomeration economies also benefit from collaboration within the industry, and that collaboration will be emphasized in an AI-powered economy. They also provide downside protection; for example, the core Kendall Square lab market posted a vacancy rate of 12.5% in Q4 2025, which is one-third the Boston metro average.

3. Target Tenant Mix Through a Sector Lens

AI will allow some to flourish while others struggle across industries. Tenants with high AI-driven risk to employment should be evaluated similarly to how work-from-home exposure was assessed in 2020—with additional risk assigned to lease expirations. Investors should incorporate AI-related downside risk into underwriting by differentiating not only the industry but also the specific functions a tenant performs in the space, especially in markets with stagnant or declining demand.

For Occupiers

1. Build Flexibility into Both Space and Leases

Hybrid work is now standard for many knowledge workers, and most organizations have adapted their workplace strategy and portfolios to this once uncertain model. Still, a little under half of pre-pandemic leases have yet to expire and these footprints will continue to adapt to today’s work patterns and preferences. As AI reshapes headcount, task mix and work patterns in difficult to anticipate ways, occupiers with upcoming lease expirations should prioritize modular, easily reconfigurable layouts and greater lease flexibility, including shorter terms, expansion and contraction options, and the use of flex space. For early-stage AI companies in San Francisco, demand has been strong for coworking or turn-key space that these companies can move into fast and have options to expand as their business grows.

2. Design Connection‑First Offices in the Age of AI

AI will likely reduce routine daily human interactions in the workplace as more workers increasingly collaborate with technology. An organization’s current office‑use patterns and workplace design should accommodate that evolution while maintaining trust, connection and culture. Ensuring that outcome depends on leadership and the physical environment. Establish the culture first, then make sure workplace design reinforces it by creating systems and environments that naturally draw people in. In an AI‑enabled workplace, time in the office will matter less for routine, heads‑down tasks and more as a platform for higher‑value collaboration, mentoring, training, problem‑solving and relationship‑building. This shifts the focus toward in-person moments rather than mandates and compliance. Office design should support these behaviors by reducing desk density and increasing high‑quality collaboration spaces, hospitality‑driven multipurpose rooms and casual “collision” zones.

3. Align Real Estate Strategy with Workforce and Automation Strategy

Corporate real estate planners should align lease and portfolio decisions with their firm’s AI and workforce transformation roadmaps. Real estate, HR and automation leaders should partner to ensure lease timing supports planned automation waves, reskilling initiatives and role redesign. AI-enabled workforce planning tools can help identify which roles are most exposed, where they are located and where flexibility should be prioritized.