On a Tuesday in March, Marc Holliday watched an AI company double its footprint at One Madison. Harvey AI, a provider of generative AI tools for law firms, expanded from 15,000 square feet to 30,000. Holliday, chairman and CEO of SL Green, called the expansion “the ultimate response to the false narrative that AI is shrinking the workforce in New York City.”
Last week, that narrative took another hit. Durst and the Port Authority announced that Norm AI, a legal and compliance platform, leased most of the top floors at 1 World Trade Center. The deal pushed the 3.1 million-square-foot tower to 97% leased. Norm AI joins tenants like BNP Paribas, the Port Authority itself, and the advertising agency GroupM.
Then came the blockbuster. San Francisco-based Anthropic, creator of the Claude chatbot, is close to a deal for the entire 365,000 square feet at 330 Hudson Street, per the Commercial Observer. Anthropic currently occupies just 15,500 square feet in the city. The lease would represent a 23x expansion in its Manhattan footprint.
Anthropic’s pending deal is not an outlier. It is the latest data point in a pattern: AI companies are absorbing large blocks of Class A office space at a moment when many predicted they would be shedding it. The fear—that AI would trigger mass layoffs and a corresponding collapse in office demand—has not materialized in the leasing data.
The mechanism is straightforward. AI companies are hiring aggressively, not firing. Anthropic raised $3.5 billion in its most recent funding round at a $61.5 billion valuation. Harvey AI raised $100 million in Series C funding in 2025. These companies need space for engineers, product managers, sales teams, and compliance officers. They are not virtual entities. They occupy desks.
Landlords are responding. SL Green’s Holliday has been vocal about the AI tailwind. The REIT’s portfolio, concentrated in Midtown Manhattan, has seen leasing velocity pick up from technology tenants. Durst’s 1 World Trade Center is effectively full. AEW Capital Management, owner of 330 Hudson, is poised to land a single tenant for an entire building.
The supply side reinforces the demand story. Manhattan office vacancy, which peaked at 22.5% in early 2024, has declined to 19.8% as of Q1 2026, per CBRE. Net absorption turned positive in 2025 and has accelerated in 2026. AI tenants are not the only driver—financial services and law firms continue to lease—but they are the marginal buyer in a market that was written off two years ago.
The capital markets implications are clear. Office REITs that own well-located, modern assets are seeing leasing momentum that supports rent growth and valuation stabilization. SL Green’s stock is up 34% over the past 12 months. Durst’s 1 WTC is generating cash flow at near-full occupancy. AEW’s 330 Hudson will go from a multi-tenant risk profile to a single-tenant credit lease if the Anthropic deal closes.
None of this means the office market is healed. The bifurcation between trophy assets and secondary space remains extreme. Buildings that lack location, amenities, or capital investment are still bleeding tenants and facing refinancing stress. But the AI-apocalypse thesis—that AI would destroy office demand—has been falsified by the data.
The lesson for investors is to distinguish between technology’s impact on labor markets and its impact on real estate demand. AI may eventually displace workers, but it is currently displacing leases. The companies building the technology need physical space to do it. That is a bullish signal for Manhattan office landlords with the right assets.
Marc Holliday’s March comment now reads as prescient. The AI narrative in commercial real estate has flipped from existential threat to demand driver. The question for the next cycle is whether that demand is durable or whether it will fade as AI companies mature and consolidate. For now, the market is voting with square feet.