The AI Office Paradox: Growth Driver, Space Reducer, or Both?

Few forces are reshaping the office landscape as quickly, or as unevenly, as artificial intelligence. It is fueling some of the strongest leasing activity the sector has seen since the pandemic, while raising real questions about what automation could mean for office space demand looking ahead.

To understand how this paradox is playing out on the ground, we asked Kidder Mathews brokers across our Western U.S. markets what they’re hearing from clients and seeing in negotiations. Rather than broad forecasting, they offered practical insight: real transaction activity, tenant behavior, and market-specific trends observed firsthand.

What emerges is a market that is anything but uniform. In talent-dense hubs like the Puget Sound, AI companies, from established players to early-stage startups, are absorbing space at both ends of the quality spectrum from trophy Class A towers to move-in-ready Class B/C suites. In San Diego, AI’s office footprint remains limited, anchored by traditional industries, and meanwhile in Phoenix, an early wave of AI-related leasing is beginning to influence net absorption.

Taken together, these perspectives reveal a common thread: AI’s impact on office demand isn’t a single story, but many, unfolding at different speeds, in different submarkets, and among different tenant profiles. The snapshots below offer a ground-level view of what is taking shape, and what it may mean for the future of office space.

Key Takeaways

• AI-driven office demand is concentrated in the Bay Area and the Puget Sound region, with Bellevue, San Francisco, and Oakland showing the clearest leasing impact.
• Demand is bifurcated by tenant type: well-funded AI firms compete for trophy Class A space with strong amenities, while early-stage startups favor move-in-ready Class B/C suites with flexible terms.
• San Diego remains largely unaffected by AI-driven office demand, anchored instead by professional office, life science, and military and defense tenants.
• Brokers across markets expect an accelerating flight to quality, with AI activity increasingly concentrated in a smaller number of high-performing Class A buildings over the next three to five years.

Seattle Market Snapshot

Kidder Mathews SVP Rod Keefe’s read on downtown Seattle centers on current leasing activity, not speculation. He’s seeing a measurable impact from AI tenants on the greater downtown office market, evidenced by a long list of startup-type groups touring and signing for their first or second office space.

That demand lands in two distinct pockets, each serving a different type of AI tenant. “On one end, Class A buildings with full amenity packages are winning over AI companies looking to compete for talent, space that helps them recruit and retain employees in a tight hiring environment,” Keefe says.

On the other end, he points to an unexpected beneficiary: older Class B/C properties with market-ready, fully furnished suites. “These spaces offer flexible lease terms and immediate move-in readiness, which matters enormously to early-stage AI companies that need to get a team into a room quickly without committing to a long build-out timeline or lease term. This bifurcated demand pattern reflects the range of AI tenants in the market, from well-funded firms prioritizing image and amenities, to leaner startups prioritizing speed and flexibility,” he explains.

Keefe sees this activity concentrated, splitting along submarket lines. Just east of Seattle, Bellevue captures larger, more established AI firms and bigger lease transactions, while downtown Seattle mostly absorbs smaller, startup-driven deals consistent with the “first or second space” tenant profile. As he puts it, “the type of AI tenant a submarket attracts is shaping the type of leasing activity that submarket sees, with company maturity and capital position acting as the dividing line between Bellevue and downtown Seattle.”

Bellevue and the Eastside Market Snapshot

Downtown Bellevue is emerging as one of the Puget Sound office market’s strongest bright spots, driven almost entirely by AI-sector expansion. Kidder Mathews EVP Gary Guenther, points to OpenAI, xAI, and CoreWeave, which together accounted for more than 350,000 square feet of leasing in the past year, with OpenAI alone occupying nearly 300,000 square feet and capacity for more than 1,000 employees. A growing bench of smaller AI startups is also competing for engineering talent. While this hasn’t solved the market’s broader vacancy challenges, Guenther calls it “one of the most meaningful sources of Class A demand since the pandemic.”

That demand, however, is highly selective. The clear winners are Class A and trophy properties in downtown Bellevue, buildings with efficient floor plates, strong power infrastructure, high-end amenities, and proximity to housing and transit. As AI companies compete aggressively for top-tier talent, the office itself has become a recruiting tool, not just a workspace. Older suburban commodity buildings, particularly those lacking amenities or well-capitalized ownership, continue to struggle to attract this tenant category.

The recovery, Guenther notes, has taken on an uneven nature. “AI-driven demand isn’t lifting the broader market, it’s concentrated where the talent is,” he says. “Bellevue and the Eastside benefit specifically from their proximity to Microsoft, Amazon, Meta, and Google outposts, making the submarket uniquely positioned to capture this activity.”

He also sees a flip side to AI’s growth story. While AI fuels demand at the top of the market, he expects it to “simultaneously erode demand for back-office functions,” adding pressure to already-struggling, equity-challenged commodity buildings.

Looking ahead five years, Guenther expects an accelerating “flight to quality,” with AI-driven growth concentrating in a smaller number of high-performing Class A buildings, particularly in innovation-oriented submarkets like Bellevue, leaving the rest of the market further behind.

Beyond established players like OpenAI, xAI, and SpaceXAI, Kidder Mathews SVP Aaron Kraft points to a growing wave of AI-driven startups translating into real leasing momentum in the submarket.

On the tenant side, Kraft doesn’t see AI companies reinventing the office wholesale. “Space usage still varies firm to firm, but for most occupiers, the practical effect of AI is productivity at the desk, employees getting more done in the same footprint, rather than a fundamental shift in how much space or what kind of space they need,” he says.

That nuance carries into his outlook on demand. Kraft is skeptical that AI-driven efficiency will hollow out office markets through layoffs. “If vacancy does rise,” he argues, “it will be a story of repositioning, property owners finding higher-value uses for the space, whether through cheaper conversion economics or improved power infrastructure, not a demand collapse driven by an AI-fueled economic downturn.”

He’s direct about where the workplace conversation stands: the return-to-office debate is over. In his view, “current office usage simply reflects company culture. Organizations that value in-person work are already back, and those that don’t have already made that choice. AI isn’t materially moving that needle either way.”

Looking five years out, Kraft’s biggest bet is on a shift in how people interact with technology at work, moving from mouse-and-keyboard input toward voice. That transition, he says, “will reshape office design on two fronts: greater demand for private workspaces to accommodate voice-based work, and more interactive, flexible spaces for employees when they step away from their desks.”

East Bay Oakland Market Snapshot

The East Bay is emerging as a key beneficiary of the AI industry’s build-out, as companies stretch beyond San Francisco for advanced manufacturing and R&D space. Oakland and Berkeley, which carried elevated post-pandemic vacancy in industrial, R&D, and office product, are seeing inventory absorbed quickly as AI and robotics firms move from proof-of-concept to execution and need high-power, build-ready facilities. West Oakland’s American Steel Blocks campus illustrates the shift: the 440,000-square-foot site was only 45% leased six months ago and is now roughly 95% full. Recent commitments there include a 100,000-square-foot lease from Jeff Bezos’ stealth AI startup, Project Prometheus, and a 40,000-square-foot lease from Tesla, which is also retooling a Fremont production line for its Optimus humanoid robot. The pattern extends across the region: OpenAI leased space in Richmond reportedly for humanoid robotics manufacturing, Ambi Robotics moved into Berkeley, and Formic Technologies took space in Oakland. This is industrial, R&D, and advanced-manufacturing demand rather than traditional office absorption, though it still shows how broadly AI is reshaping real estate across property types.

For landlords, the near-term effect is a meaningful inventory tightening in a submarket that had struggled since the pandemic; American Steel Blocks’ jump from 45% to 95% leased in six months is a striking data point on its own. Brokers describe it as more than a one-off: once a few high-profile names commit, it lowers the perceived risk for others considering the same move, creating a self-reinforcing draw. Whether that translates into sustained, long-term absorption across the wider East Bay market will hinge on how many AI firms progress from funded startups into full-scale manufacturing, but the early signal points to continued—not one-time—demand.

San Francisco Market Snapshot

AI companies have become the primary driver for office demand in San Francisco, fueling a resurgence in leasing activity across key submarkets. Established firms like Open AI and Anthropic have each leased more than one million square feet over the past two years and are now two of the city’s top five tenants, supporting rapid hiring with a preference for high-quality buildings, strong amenities, and talent-friendly locations. Early-stage AI startups are also active, seeking smaller, move-in-ready spaces with flexible layouts and room to expand, even as landlords push for longer lease commitments amid tightening availability. This broad-based demand, concentrated in San Francisco’s leading innovation districts, is helping absorb available inventory and reinforcing the city’s position as a premier AI hub.

Los Angeles Market Snapshot

Los Angeles has emerged as a leading, applied-AI growth market, supported by an expanding talent pool and evolving innovation ecosystem, with companies building solutions for industries central to the Southern California economy, including entertainment, media, healthcare, aerospace, and defense. Much of this activity is concentrated in hubs like Santa Monica, Playa Vista, Culver City, El Segundo, and Pasadena, attracting both established companies and startups. As investment and hiring grow, AI companies are expected to further expand their footprint, reinforcing Los Angeles’ position as a leading innovation center and growing source of office demand.

San Diego Market Snapshot

Unlike other West Coast markets, San Diego hasn’t emerged as an AI office-demand hub. Kidder Mathews VP Jeff Gilbert says the region remains anchored by its traditional strengths, professional office, life science, and military/defense, rather than a concentrated AI tenant base. San Diego has historically produced innovative companies, he notes, but its smaller tech firms tend to get acquired and absorbed elsewhere rather than growing into major local occupiers, while San Francisco, Silicon Valley, and the Pacific Northwest capture the bulk of AI-driven demand thanks to deeper talent pools.

Where AI activity touches San Diego, it mirrors national patterns: large, well-capitalized tech companies drive it, gravitating toward Class A properties with top-tier amenities to win the talent war. But Gilbert says, “this is a selective, market-specific phenomenon, AI-driven demand is concentrated almost entirely in the Bay Area and Seattle/Pacific Northwest, while San Diego’s broader office market remains flat, dominated by traditional users rather than AI-driven growth.”

Gilbert sees a few consistent patterns in how AI tenants differ: stronger in-office attendance expectations than typical hybrid norms, space built around dense collaboration rather than individual workstations, faster and less predictable growth that favors flexible lease terms, and outsized emphasis on amenities and ‘destination’ quality for recruiting. He describes these tenants as “behaving more like high-growth, capital-rich startups than mature enterprise users,” adding that “AI companies are actively pulling engineers and coders back into the office to move faster and collaborate more closely.”

Looking further out, Gilbert believes automation will eventually create real headwinds for office demand—not immediately, but over the medium term—as businesses find efficiencies that reduce headcount in certain sectors. He notes this could also spawn new employment categories elsewhere, softening the net impact.

Phoenix Market Snapshot

Kidder Mathews FVP Kurt Kerner sees Phoenix in the early innings of an AI-driven demand wave, with new companies entering the market, raising capital, hiring aggressively, and signing office leases. That activity has been a meaningful contributor to the market’s positive net absorption this year.

Kerner expects a more nuanced picture over the next three to five years. “As AI acceptance achieves mainstream adoption across industries, businesses are broadly expected to become more efficient, with some reducing overhead, streamlining operations, and operating with leaner headcounts as a result,” he says.

Still, Kerner isn’t bearish long term. He frames this historically: “technology has consistently displaced certain jobs while simultaneously creating entirely new industries.” He expects a second wave to follow, as the workforce adapts, builds new skills, and forms businesses around technologies that don’t yet exist. Those companies will, in turn, need office space of their own.

His bottom line: “AI will reshape who occupies office space and how much they need at different points in time, but it won’t fundamentally change the underlying, long-term need for people to collaborate and build businesses together in person.”

AI-Related Office Leasing Activity

Figures as reported by Kidder Mathews brokers.

The Bottom Line

Together, these perspectives make clear that AI’s influence on Western U.S. office markets defies a single narrative: in some regions, it’s already a measurable demand driver, absorbing space at both the trophy and value ends of the market; in others, that story is just beginning, with early-stage leasing starting to influence net absorption; elsewhere, the picture is more muted, a reminder that proximity to deep technical talent remains a prerequisite for AI-driven demand, not a guarantee.

What unites these markets isn’t the presence or absence of AI tenants, but a shared expectation that flight to quality will keep accelerating with Class A, amenity-rich, talent-competitive buildings capturing outsized demand regardless of sector. Whether AI expands the office footprint or compresses it through efficiency gains, brokers agree: the office itself, location, quality, and ability to support in-person collaboration, remain central to how companies compete for talent in an AI-driven economy.