What is AI office job displacement risk? AI office job displacement risk is the share of office-using employment that artificial intelligence could automate, and by extension the amount of occupied office space that could disappear along with those jobs. On September 9, 2026, CBRE Research put a hard number on it in AI's Impact on Office Demand: just 5% of the US office-using workforce is considered highly vulnerable to AI displacement. For the wider context on how these tools are reshaping the sector, see our guide to AI commercial real estate.
Key Takeaways
- CBRE Research finds only 5% of the US office-using workforce is highly vulnerable to AI displacement, while 18% is highly adaptive and 77% sits in between.
- Even if every at-risk office job vanished, CBRE projects US office vacancy would reach just 18.7% by 2031 versus 18.3% today, a 40 basis point swing.
- CBRE's baseline forecast has office vacancy falling to 14.5% by year-end 2031, a 380 basis point improvement over today's level.
- The jobs most exposed to automation skew heavily remote, so displacing them removes payroll cost but very little occupied square footage.
- The real constraint on office-using job growth is demographic, with roughly 83,000 workers retiring monthly and sharply reduced immigration.
What CBRE's September 2026 Analysis Actually Found
CBRE segmented the entire US office-using workforce into three tiers of AI exposure. Only 5% is highly vulnerable to displacement. Another 18% is highly adaptive, meaning AI makes those workers materially more productive. The remaining 77% is exposed to change but not to elimination: the tasks shift, the headcount largely does not.
The sector detail matters more than the headline. Finance, professional services, and administrative services employ the largest raw number of vulnerable workers, which is why AI job-loss stories so often feature banks and consultancies. But just 6% of that combined workforce is actually exposed. The concentration reflects those sectors being enormous, not uniquely fragile.
The roles CBRE flags as genuinely at risk are narrow and clerical, including payroll and timekeeping clerks and tax collectors and revenue agents. The roles it identifies as AI-supported are precisely the ones that anchor a lease: financial and investment analysts, computer network architects, and AI engineers. The report is the second in a four-part CBRE Research series on AI and commercial real estate.
The Stress Test That Caps the Bear Case
The most useful number in the report is not the 5%. It is what CBRE calls its AI Vulnerability Downside Scenario, which assumes every single at-risk office-using job is eliminated. Under that deliberately severe assumption, US office vacancy rises from today's 18.3% to 18.7% by 2031. That is 40 basis points, a rounding error against a sector that has repriced by hundreds of basis points on this fear.
Set that against CBRE's baseline, in which vacancy declines to 14.5% by year-end 2031, a 380 basis point improvement. The distance between the base case and the apocalypse case is roughly one tenth the distance between today and the base case. For anyone underwriting office, that ratio is the finding. The variables that actually move office NOI over the next five years are supply, capital costs, and tenant credit, not the automation rate of clerical work.
CBRE also notes that private sector layoffs are running near their lowest rate since 2013, which is hard to square with a narrative of AI-driven white collar collapse. Our earlier analysis of whether AI will kill commercial real estate laid out the bear case in full; this dataset is the first credible attempt to bound it.
Why the Jobs AI Displaces Were Not in the Office Anyway
The mechanism is the part most coverage misses, and it is what separates this report from another leasing forecast. CBRE's framing is blunt: the jobs AI threatens have mostly left the office already, and the jobs AI supports favor in-office work.
Highly automatable roles such as data entry, claims processing, and routine bookkeeping are disproportionately performed remotely. They were among the first to go fully distributed after 2020 and the last to be called back. Eliminating them removes payroll expense from a tenant's income statement but removes very little occupied square footage from a landlord's rent roll.
The inverse holds too. The work AI struggles to replace is complex, judgment-intensive, and collaborative: negotiation, supervision, client development, and training junior staff. That work skews strongly toward in-person environments. So the composition shift runs in the landlord's favor, thinning the remote-performed tail of the workforce while concentrating survivors in roles with a genuine reason to occupy space. CBRE invokes the Jevons Paradox here: as AI makes professional services cheaper and faster, aggregate demand for those services rises, a dynamic already visible in surging business formation rates. CBRE also points out that office-using employment posted its largest gains in share of total US jobs after both the internet and the smartphone arrived.
What This Changes in Office Underwriting
Three practical adjustments follow for investors and operators.
- Stop treating an AI vacancy shock as a base case. If the worst credible scenario is 40 basis points of vacancy, an AI-specific haircut on office cash flows is not defensible as a headline assumption. Model it as a sensitivity instead.
- Underwrite the tenant's job mix, not its industry label. A tenant whose floor is clerical processing carries different renewal risk than one whose floor is analysts and engineers, even inside the same industry classification. That distinction belongs in tenant credit review alongside lease term, guarantor strength, and renewal probability, and it flows through to the NOI durability your DSCR ultimately depends on.
- Concentrate on quality and workforce depth. CBRE finds prime office and top-quality assets best positioned, and markets with highly educated, adaptable workforces facing the least AI-related risk. This reinforces rather than reverses the flight-to-quality trade.
CRE investors looking for hands-on help translating findings like these into an actual underwriting model can reach out to Avi Hacker, J.D. at The AI Consulting Network.
The Caveats Before You Reprice Anything
This is a forecast, and forecasts about technology adoption have a poor track record in both directions. CBRE's vulnerability tiers rest on task-level assumptions about what current models can do. A genuine capability jump in agentic systems from OpenAI, Anthropic, or Google could move workers between tiers faster than the model anticipates.
National averages also conceal market-level pain. A 40 basis point national swing can still contain a metro where one clerical-heavy employer vacates 300,000 square feet and resets comparable rents for a submarket.
Finally, CBRE is a brokerage with a commercial interest in office optimism, so the direction of its institutional bias is worth naming even where the methodology looks sound. Read it alongside the demand-side evidence in our analysis of AI tenants as an office demand engine and CBRE's separate talent data on which sector is now hiring AI workers. Trade coverage of the release is available via Connect CRE.
Frequently Asked Questions
Q: How many office jobs are actually at risk from AI?
A: CBRE Research found that just 5% of the US office-using workforce is highly vulnerable to AI displacement, with 18% highly adaptive and the remaining 77% exposed to task changes rather than job loss.
Q: Will AI increase office vacancy rates?
A: CBRE's downside scenario, which assumes every at-risk office job is eliminated, projects vacancy of 18.7% by 2031 against 18.3% today. Its baseline forecast instead has vacancy falling to 14.5% by year-end 2031.
Q: Why does AI job displacement not reduce office demand proportionally?
A: The most automatable roles are disproportionately performed remotely, so eliminating them cuts payroll without vacating space. The roles AI supports tend to be collaborative, judgment-intensive, and office-based.
Q: What is actually constraining office-using job growth?
A: Demographics rather than automation. CBRE points to roughly 83,000 workers retiring per month plus sharply reduced immigration, and forecasts 0.9% annual office-using job growth, still ahead of 0.6% for the broader economy.
Reading a research release is easy; changing an underwriting model because of it is harder. If you want help deciding which assumptions in your office model deserve to move, The AI Consulting Network specializes in exactly this.