The technology displacing workers is also, for now, driving one of Manhattan’s strongest leasing recoveries in a decade.
AI firms signed more than 100 leases across Manhattan in 2025 — a 60% jump from the prior year — adding roughly 1 million square feet of office space. Legacy tech firms investing in their own AI capabilities added another 2.1 million. The result: Manhattan’s best leasing year since 2014.
For commercial real estate professionals, the numbers are a welcome reprieve after five years of elevated vacancies and tepid demand from the tech sector. But the boom rests on an uncomfortable paradox. The companies driving this office revival are building tools explicitly designed to reduce reliance on human labor — and the market knows it.
For corporate real estate portfolios, this is a structural market shift with direct implications for supply, pricing power, and long-term space strategy.
The Numbers: A Market in Transition
The scale of AI-driven leasing activity is hard to overstate. Consider the activity across Manhattan alone in 2025:
| Metric | Figure | Context |
| AI firm leases signed | 100+ | +60% YoY |
| AI square footage added | ~1M sq ft | +152% from 2024 |
| AI sq ft currently sought | 1.4M sq ft | Active in market |
| Legacy tech sq ft added | 2.1M sq ft | AI-driven expansion |
| Average AI rent PSF | $88 | vs. $78 citywide avg |
| Peak AI deal | $210 PSF | Select trophy assets |
| Tech share of top 20 leases | ~33% | Up from ~10% prior year |
Headline deals include Harvey AI absorbing more than 185,000 square feet at One Madison Avenue — fully occupying a building that had struggled since its 2023 redevelopment — and OpenAI leasing roughly 90,000 square feet in SoHo for its first New York office.
Anthropic is actively seeking to expand its Manhattan footprint. Elise AI is moving into the former Tiffany building with a 50%-plus expansion. Even Palantir, whose co-founder made headlines for his sharp criticism of Mayor Zohran Mamdani, is reportedly seeking additional space in the city.
“Every industry and company is thinking about how to implement AI technology, and New York is ground zero for that conversation. And every company is here. — Julie Samuels, CEO, Tech:NYC”
Why New York — and Why Now
New York is the second-largest tech hub behind the Bay Area, but several structural shifts have accelerated its AI moment:
- NYC accounts for more than 9% of the country’s AI workforce, leading Seattle, Boston, and Los Angeles.
- Overall tech employment across the five boroughs grew 12% from 2020 to 2024, with an additional 13% projected by 2029 (JLL Research).
- The number of tech firms in Manhattan rose 21% from 2020 to 2024.
- New York boasts more than 8,750 startups — more than San Francisco — and posted more than 25,000 AI-related job openings in 2025, a national high (Tech:NYC / Center for an Urban Future).
- San Francisco remains crowded and expensive; New York offers access to capital, financial sector clients, and a deeper talent pool across verticals.
The cascading effect on submarkets is already visible. As premier Midtown addresses tighten, AI tenants are pushing into previously overlooked districts.
Scale AI relocated from Chelsea to the Financial District after its headcount doubled to 500. Topline Pro, an AI-powered platform for home services companies, chose a converted industrial building in Williamsburg, Brooklyn — deliberately positioning near where employees live and requiring five days per week in-office. Ramp, the financial tech firm, is adding two floors near Madison Square Park.
The Counterforce: AI as an Office Demand Destroyer
For Fortune 1000 occupiers, the current narrative deserves scrutiny. The same firms leasing space today are explicitly building tools designed to reduce human labor. That tension is already visible in equity markets.
Shares of SL Green Realty Corp. and Vornado Realty Trust, two of Manhattan’s largest office landlords, have declined in 2026 in part due to investor concern about AI’s long-term impact on office demand. The market is pricing in a scenario where today’s leasing cycle is real but finite — and where the efficiency gains driving AI growth ultimately compress the headcount that justifies corporate footprints.
For large occupiers managing multi-million-square-foot portfolios across multiple markets, the risk calculus is significant:
- Workforce compression from AI automation may reduce long-term headcount — and with it, the space required to house it
- Lease commitments signed today at premium rents may not align with headcount 5-7 years forward
- As AI tenants drive up asking rents in secondary submarkets, large occupiers face rising renewal and relocation costs
- Landlords are leveraging tight conditions to push longer terms and higher TI structures, increasing portfolio inflexibility
- Sectors most exposed to AI disruption — legal, financial services, professional services — are also the heaviest Manhattan office users
The irony is pointed: corporations deploying AI to reduce costs may simultaneously be absorbing higher occupancy costs driven by AI companies hiring aggressively. Both forces are operating in parallel, and the net effect on portfolio strategy remains unresolved.
Pricing Pressure: What the Data Says About Rents
The tightening is translating directly into rent escalation — particularly in the trophy and near-trophy segment. As of 2025:
- Citywide average asking rent: $78 per square foot
- AI firm average rent paid: $88 per square foot — a 13% premium over market
- High-water mark in AI deals: $210 per square foot
- Most sought-after addresses: $300+ per square foot
- Venture-backed real estate in AI corridors (Little Italy, SoHo, Flatiron): reported 40% rent increases since 2022 in some buildings
For tenants approaching lease expirations in high-demand submarkets, these figures are not abstract. Landlords with leverage will push — and most tenants are negotiating without complete visibility into what comparable deals are actually closing at, not just what is being marketed.
What This Means for Corporate Occupiers
The current market creates a specific challenge for large corporate tenants: conditions are tightening faster than most portfolio strategies anticipated, and the data asymmetry between landlords and tenants has never been greater.
Landlords and their brokers have access to full transaction data across every comparable deal in a submarket. Most tenants — even those with sophisticated in-house CRE teams — are working from marketed asking rents, anecdotal comps, and broker-curated comparables. That gap has a dollar value, and in the current environment, it is substantial.
The most important questions for Fortune 1000 occupiers in this market are not strategic. They are transactional:
- What did the comparable tenant in the same building pay — not what was listed, but what was executed?
- What free rent, TI allowances, and concession structures are landlords actually granting in this submarket right now?
- Is the proposed rent above or below where deals are actually clearing in this asset class and location?
- What does the forward supply pipeline look like in 18–36 months — and does it shift leverage?
The difference between knowing what tenants should pay and knowing what they are paying is the difference between a market-rate deal and an above-market one. In the current environment, the gap can represent millions of dollars over a lease term.
STOP NEGOTIATING BLIND
Know What You Should Be Paying — Before You Sign
REoptimizer® is the first CRE transaction management platform built specifically to close the data gap between tenants and landlords. Powered by AI and trained on over 8,000 data points per transaction, REoptimizer® doesn’t tell you what you should pay — it tells you what you are paying relative to what deals are actually closing at, right now, in your market.
In a market where AI companies are setting new rent ceilings and landlords have more leverage than they have in a decade, data parity isn’t a nice-to-have. It’s a negotiating requirement.
