OpenAI Takes Full 350,000 SF Mountain View Campus in Major Expansion

OpenAI has leased the entire 350,000-square-foot office campus at 350–380 Ellis Street in Mountain View, California, in a deal that underscores the AI company's continued aggressive expansion in the Bay Area despite broader uncertainty in commercial real estate.

The campus, owned by KKR Real Estate Finance Trust Inc. (KREF) and TMG Partners, comprises two Class A office buildings totaling roughly 350,000 square feet. The lease was announced March 23, 2026, though financial terms weren't disclosed. KREF and TMG didn't specify lease duration or rental rates, but comparable recent deals in Mountain View's North Bayshore submarket have commanded $5.50 to $7.00 per square foot monthly for renovated space.

The deal comes as OpenAI—valued at $157 billion in its October 2024 funding round—continues adding physical footprint across the San Francisco Bay Area. The company already occupies significant space in San Francisco's Mission District and has been quietly expanding into additional buildings as headcount grows. This Mountain View lease represents OpenAI's largest single campus commitment to date and its first major presence outside San Francisco proper.

"We're thrilled to welcome OpenAI to this premier campus," said Michael Covarrubias, CEO of TMG Partners, in the announcement. What he didn't say: this lease solves a vacancy problem that's plagued the property since its previous anchor tenant, a legacy tech firm, vacated most of the space in 2023. The Ellis Street campus sat roughly 60% empty for nearly two years before OpenAI entered negotiations last fall.

Mountain View isn't an obvious choice for a company that's built its identity around San Francisco. But the Ellis Street campus offers something OpenAI's fragmented SF footprint doesn't: consolidated space under one roof, immediate availability, and proximity to both Stanford talent pipelines and the infrastructure-heavy data center corridor stretching south toward San Jose.

The campus sits less than two miles from Google's main campus and a short drive from the Caltrain station, making it accessible for employees commuting from San Francisco. More importantly, it's positioned in a submarket where vacancy rates—while elevated at 18.3% as of Q4 2025—are stabilizing faster than San Francisco's downtown core, which remains above 30%.

OpenAI's headcount has reportedly crossed 5,000 employees globally, up from roughly 1,500 in early 2023. The company has been hiring aggressively across engineering, research, and go-to-market functions as it races to maintain its lead in generative AI. A single 350,000-square-foot campus provides room to house 1,500 to 2,000 workers depending on density—enough to consolidate several scattered offices or accommodate a full department.

There's also a competitive angle. Anthropic, OpenAI's chief rival, recently expanded into additional space in San Francisco's SoMa district. Google DeepMind occupies substantial space across Mountain View and London. Consolidating teams under one roof isn't just operational efficiency—it's a signal to employees and competitors that the company isn't slowing down.

For KREF and TMG Partners, landing OpenAI represents a significant win in a market where landlords have struggled to backfill space vacated during the pandemic and subsequent tech layoffs. KREF, a publicly traded real estate investment trust managed by an affiliate of Kohlberg Kravis Roberts & Co., specializes in commercial mortgage loans and equity investments. TMG Partners is a San Francisco–based real estate investment and development firm with a portfolio concentrated in Bay Area office and mixed-use properties.

The Ellis Street campus had been partially occupied by a mix of tenants since its previous anchor departed, but never fully leased. Landing a single creditworthy tenant for the entire campus eliminates leasing risk, reduces operational complexity, and provides a stable cash flow anchor for the asset.

KREF's involvement is particularly notable. The REIT has been under pressure from investors to stabilize its portfolio amid rising interest rates and concerns about office valuations. Securing a long-term lease with a well-capitalized tenant like OpenAI provides a concrete answer to questions about whether premium Bay Area office assets can still command institutional interest.

"This lease demonstrates the continued demand for high-quality office space in strategic locations," said Matt Salem, President and Chief Investment Officer of KREF, in the joint announcement. Translation: this deal proves that reports of the office market's death have been exaggerated—at least for the right tenant in the right building.

The broader Bay Area office market remains deeply bifurcated. Premium, amenity-rich campuses with good transit access are seeing selective leasing activity from AI companies, life sciences firms, and well-funded tech players. Meanwhile, older Class B and C buildings—especially in San Francisco's Financial District—continue to hemorrhage tenants.

Mountain View's North Bayshore submarket, where the Ellis Street campus sits, has seen net positive absorption over the past two quarters, driven largely by AI and infrastructure-related companies. This stands in contrast to San Francisco's downtown core, which posted negative absorption for the eighth consecutive quarter in Q4 2025.

Recent comparable leases in the Mountain View area provide context for the deal's significance:

OpenAI's lease ranks among the largest new commitments in the submarket over the past 18 months. It's also one of the few instances where a tenant took an entire multi-building campus rather than leasing a portion of a larger complex.

This deal reinforces what brokers and landlords have been arguing for two years: the office market isn't dead, it's bifurcating. Companies with capital and growth trajectories still want physical space—they're just far more selective about location, building quality, and amenities than they were pre-pandemic.

The Mountain View lease fits into a pattern of rapid, sometimes chaotic expansion. OpenAI has added space in fits and starts over the past three years, often signing short-term or flexible leases to accommodate unpredictable headcount growth. The company's original office at 3180 18th Street in San Francisco's Mission District has long since overflowed.

In 2024, OpenAI reportedly added multiple floors in a building near the Embarcadero and began leasing space in a mixed-use development in the Dogpatch neighborhood. Those moves were piecemeal—grabbing available space as it came to market rather than executing a long-term real estate plan.

The Ellis Street campus, by contrast, looks like strategic infrastructure. It's large enough to house a full department—research, product, infrastructure engineering—rather than scattering teams across buildings. And it's close enough to San Francisco for executives to commute but far enough from the city's operational chaos to offer employees a different environment.

There's also a talent retention angle. Some employees, particularly those with families or longer commutes, have pushed back against San Francisco–only expansion. A South Bay presence gives OpenAI access to talent pools that wouldn't otherwise consider a city-based role.

The company hasn't publicly detailed how it plans to use the space—whether it'll serve as a research hub, an engineering center, or a consolidation point for scattered teams. Requests for comment from OpenAI went unanswered before publication.

OpenAI's decision to expand in Mountain View rather than taking additional space in San Francisco won't help the city's struggling office landlords. While the company isn't abandoning its Mission District headquarters, the Mountain View lease signals that future growth may happen outside city limits.

San Francisco's office market has been waiting for AI companies to absorb vacant space and reverse negative absorption trends. Instead, those companies are proving more selective than anticipated—choosing newer buildings, suburban campuses, or markets with lower costs and better access to infrastructure.

It's tempting to read this lease as evidence that office demand is stabilizing. But zoom out and the picture is less rosy. OpenAI is an outlier: a company with $13 billion in funding, explosive revenue growth, and a hiring trajectory that defies broader tech trends.

Most tech companies are still shrinking their footprints. Meta has handed back over a million square feet of Bay Area space since 2022. Salesforce continues to sublease floors in its San Francisco tower. Amazon paused or canceled multiple office projects. The fact that landlords are celebrating a single 350,000-square-foot lease reveals how scarce large deals have become.

And even for OpenAI, questions remain. The company's valuation is tied to expectations of sustained revenue growth and dominance in an increasingly crowded AI market. If competition intensifies or revenue growth slows, OpenAI could find itself with more space than it needs—just like the tech companies that overexpanded in 2021.

For now, though, the company is betting that its trajectory justifies the commitment. And KREF and TMG Partners are betting that OpenAI's lease will hold through whatever volatility lies ahead.

KREF and TMG Partners are taking on exposure to a single tenant across an entire asset. If OpenAI's business falters or the company decides to consolidate operations elsewhere, the campus could go from fully leased to fully vacant overnight. That's a risk diversified landlords typically avoid—but in today's market, securing any creditworthy tenant for 350,000 square feet is worth the concentration risk.

TMG Partners has a long history in the Bay Area and a track record of repositioning assets through down cycles. KREF, meanwhile, has the backing of KKR and access to capital that smaller landlords lack. Both are betting they can manage the risk—and that OpenAI's growth story has years left to run.

The OpenAI lease is likely to trigger more landlord activity in Mountain View and surrounding submarkets. If one AI company will commit to a full campus, others might follow. Expect brokers to start pitching available space to Anthropic, Cohere, Midjourney, and other well-funded AI players as consolidation opportunities.

But the supply-demand imbalance remains stark. Mountain View still has millions of square feet of available space, much of it in older buildings that don't meet the standards AI companies expect. Landlords without premium assets or capital for renovations will continue struggling to compete.

The deal also raises questions about how long the AI boom can sustain this level of real estate absorption. Generative AI is capital-intensive, and most companies in the space are still burning cash. If funding conditions tighten or revenue models prove harder to scale than expected, today's aggressive expansion could become tomorrow's sublease glut.

For now, though, the narrative is clear: AI companies are the only segment of the tech industry still expanding aggressively into physical space. And landlords with the right assets in the right locations are finding that there's still demand—if you're willing to wait for the right tenant.

This lease doesn't prove that the Bay Area office market has turned a corner. It proves that a handful of exceptionally well-funded companies are still growing fast enough to need space—and that landlords are willing to wait years for those tenants rather than accept the alternative.

The tech industry's real estate footprint is consolidating around a smaller number of companies with clearer growth trajectories. OpenAI, Nvidia, and a few others are expanding. Everyone else is contracting or holding steady. The result is a market where the gap between premium assets with strong tenants and everything else continues to widen.

The Ellis Street campus deal will be held up as evidence that the worst is over. But the data suggests something more nuanced: the market is sorting itself into winners and losers, and the distance between them is growing.

OpenAI's expansion is real. The question is whether it's a leading indicator of broader recovery or an outlier that makes the underlying weakness easier to ignore.

Several developments will clarify whether this deal represents a trend or an anomaly. First, watch whether other AI companies follow OpenAI into the South Bay. If Anthropic, Cohere, or others sign similar leases in the coming quarters, it suggests a genuine shift in where AI companies want to build out infrastructure.

Second, monitor OpenAI's headcount trajectory. The company's ability to fill 350,000 square feet depends on sustained hiring. If growth slows or layoffs hit the sector, this lease could become a liability.

Third, track KREF's performance and whether this lease stabilizes investor sentiment around Bay Area office assets. If KREF's stock responds positively and other REITs start marketing similar properties to AI tenants, it signals that institutional capital sees a path forward.

Finally, watch the sublease market. If tech companies continue dumping space faster than AI firms can absorb it, vacancy rates will keep climbing regardless of high-profile deals like this one.

The OpenAI lease is significant. But one deal doesn't make a recovery. It just proves that in a deeply troubled market, there are still a few companies willing to bet big on growth—and a few landlords willing to bet big on them.