How do I responsibly make a 10-year lease commitment when our business can't see 18 months ahead?
That's not a rhetorical question. It's the actual bind CRE leaders are in right now. And it's getting worse, not better.
For decades, the assumptions underneath long-term leases were stable (or at least trusted) enough to justify the commitment. Headcount grew on a relatively predictable curve. You could model hiring, apply a seats-per-person ratio, and arrive at a square footage number with reasonable confidence. The lease matched the forecast. The forecast matched reality. Close enough, anyway.
That model is breaking down. And AI is the reason it's not coming back.
The mechanism worth understanding.
The conversation about AI and jobs tends to collapse into two camps: the catastrophists who think AI eliminates everything, and the skeptics who think the disruption is overstated. Neither framing is useful for a CRE leader trying to make a real estate decision this quarter.
Here's what's actually happening, and why it matters specifically for how you size a portfolio.
AI isn't eliminating functions wholesale, at least not yet, and not uniformly. What it's doing is changing the ratio of output-to-headcount inside functions that already exist. A legal team that needed twelve people to handle a given volume of contract review may now need eight. A finance team running FP&A may find that the analytical work that justified three analyst hires last year now requires one. A sales team may find that the research, outreach, and follow-up tasks that once required dedicated headcount are getting absorbed into the workflow of existing reps.
None of these are dramatic, headline-grabbing disruptions. They're quiet compressions. And they're happening across knowledge work broadly, in ways that don't show up as layoffs but do show up as slower hiring, smaller backfills, and a quieter reality: teams doing more without growing.
We're watching this play out in our platform data in real time. Since early 2023, average attendees per meeting room booking has declined from 3.2 to 2.6, a 20% drop in group size, while booking volume over the same period has grown by more than 40%. Teams are meeting more often. Fewer people each time. The space requirement per session is declining, not growing.
That matters when you're staring at a renewal on 50,000 square feet sized for a team that was 320 people in 2022 and is 260 people today.
You're being asked to make a 10-year bet on a forecast that has a material chance of being meaningfully wrong within the first 2.
The lease isn't the problem. Treating it as the only instrument is.
For more than 50 years, corporate real estate portfolios have been built on long-term leases. This made sense when the forecast was reliable. When you knew, within a reasonable margin, how many people you'd have in a given city over a given period and when you reliably knew that people would do most of their work at one central office location,, you could build a portfolio around these bedrock assumption and be right often enough that the model held.
Most portfolios are still built that way. And they're now carrying assumptions about headcount, utilization, and work patterns that were reasonable in 2021, questionable in 2023, and increasingly disconnected from what the data actually shows today.
This is what stranded assets look like in practice. It's not that the space is empty. It's that it's funded by a forecast that no longer matches the organization, and the cost rolls forward regardless.
Building to absorb being wrong.
The companies navigating this well aren't the ones with better forecasts. They're the ones that have reconsidered their portfolio management activities to absorb forecast error rather than pretend it doesn't exist.
That means a different mix of commitments. More fixed where you have real confidence. More flexible where you don't.
T-Mobile made this shift. They closed underutilized field offices across the country, replaced fixed commitments with policy-governed access to on-demand space, and saved 80% against their prior real estate spend in those markets. Allstate made this shift. They embraced their distributed DNA and reimagined their portfolio to match the way they work. For both these companies and many others, the portfolio didn't shrink in coverage - in fact, it grew. It just got a whole heckuva lot more efficient. And those savings are funding smarter investments across the business.
What made it work wasn't a better forecast. It was building the portfolio so it could respond when the forecast turned out to be wrong, as it always eventually does.
The question for your next renewal.
As your existing portfolio leases come up for renewal over the months and years ahead, the question isn't just whether to renew, restack, or relocate. It's whether the commitment you're considering is sized for a world where you can model your headcount with confidence, or for a world where you can't.
Right now, you're operating in the second world.
The forecast is broken. Your lease doesn't know that yet. The upcoming renewals are the moments to close that gap.
Not with a better model, but with a portfolio that suits your work, today, tomorrow, and beyond.