What is the CRE AI adoption data gap? The CRE AI adoption data gap is the disconnect between widespread AI experimentation in commercial real estate and the small share of firms that have actually deployed AI at the enterprise level, driven largely by concerns over giving outside AI vendors access to proprietary deal, tenant, and portfolio data. At Bisnow's New York AI and Technology event on May 13, 2026, Fundrise CEO Ben Miller put a sharp point on the issue from the stage at Silverstein Properties' 7 World Trade Center. For a broader view of where AI fits in CRE today, see our pillar guide on AI commercial real estate.
Key Takeaways
- Fundrise CEO Ben Miller said CRE firms "feel like they have all this special data" and are afraid to share it with Claude or OpenAI, slowing adoption.
- A Keyway survey found nearly half of real estate companies are running AI pilots, but only 9% have deployed AI at the enterprise level.
- Only 8% of CRE firms surveyed are "data-ready" for AI, meaning their data is clean, structured, and accessible enough to power AI workflows.
- More than half of surveyed firms plan to increase AI spending by over 20% in the next 24 months, signaling intent to close the gap.
- The fix is enterprise-grade AI deployment with data isolation, vendor due diligence, and a structured data-readiness roadmap.
What Ben Miller Said at Bisnow's New York AI Event
On May 13, 2026, Bisnow hosted its New York AI and Technology event at Silverstein Properties' 7 World Trade Center. Fundrise CEO Ben Miller, whose firm manages billions in retail-investor real estate capital, articulated the dilemma facing commercial real estate firms more directly than most industry voices to date. According to Bisnow's coverage, Miller said CRE firms "feel like they have all this special data. They don't want Claude or OpenAI to have access to the data... so they're afraid to adopt the cutting-edge."
That single quote captures a tension that nearly every CRE firm is wrestling with right now. Years of rent rolls, T12s, broker opinions, lease abstracts, and deal pipeline data feel like proprietary moats. Yet the same firms watch competitors automate underwriting, due diligence, asset management, and tenant communications with tools like Claude, ChatGPT, Gemini, and Perplexity, and they wonder if their caution is becoming a competitive liability. This is the same trust gap surfaced in the First American CRE broker AI trust gap survey, where 66% of brokers reported using AI daily but only 5% trusted it for actual deal decisions.
The Keyway Survey: What the Numbers Show
Proptech firm Keyway released a survey alongside the event that puts hard numbers on the trust gap. The headline findings:
- Nearly 50% of real estate companies are running AI pilots, meaning experimentation is now mainstream across the industry.
- More than 50% plan to increase AI spending by over 20% in the next 24 months, showing real budget commitment.
- Only 9% have successfully deployed AI at the enterprise level, exposing a wide gap between pilot work and production capability.
- Only 8% are "data-ready" for adoption, meaning their data quality and infrastructure can actually support AI workflows at scale.
Those four data points line up with what we see across our client base at The AI Consulting Network. CRE investors and operators are spending money on AI, but the data infrastructure to turn that spend into measurable ROI is not yet in place. The result is a pile of disconnected pilots, a few impressive demos, and very little enterprise-wide capability.
Why CRE Firms Are Afraid of AI Vendors
The data-fear narrative is not irrational. CRE firms have three legitimate concerns when they evaluate Claude, ChatGPT Enterprise, Gemini, or any frontier model vendor:
- Training data leakage: The fear that proprietary data uploaded to a model provider will be used to train future models that competitors can query.
- Counterparty risk: Concerns about a vendor outage, breach, or policy change that could expose sensitive tenant, broker, or LP data.
- Regulatory exposure: New AI compliance frameworks like the EU AI Act, the Colorado AI Act, and state-level disclosure rules raise questions about what data can legally be sent to which vendors.
The good news is that enterprise-tier offerings from Anthropic (Claude for Work), OpenAI (ChatGPT Enterprise), Google (Gemini for Workspace), and Microsoft (Copilot for Microsoft 365) all contractually commit not to train on customer data. The differences between consumer and enterprise plans matter a lot here. We covered this in depth in our guide on consumer vs enterprise AI plans for CRE.
From Pilot to Enterprise Deployment
The 9% number is the most important takeaway from the Keyway survey. The gap between running pilots and deploying at the enterprise level is where most of the wasted CRE AI spend lives. The firms that close this gap tend to follow a consistent pattern:
- Data inventory first: Catalog the data sources that matter most (T12s, rent rolls, lease abstracts, deal pipelines, market comps) before picking tools.
- Choose enterprise plans, not consumer accounts: Negotiate data-use terms, retention controls, and audit logs upfront with the vendor.
- Start with narrow, high-leverage workflows: Underwriting memo drafting, lease abstraction, OM summarization, and broker email triage are common early winners.
- Measure outcomes, not activity: Track hours saved per analyst, deals reviewed per week, and underwriting cycle time, not just license counts.
For CRE investors who want hands-on help bridging the pilot-to-enterprise gap, Avi Hacker, J.D. at The AI Consulting Network specializes in exactly this kind of structured rollout.
The Counterargument: Why Some CRE Data Fears Are Overrated
Miller's framing implies that CRE firms overrate the uniqueness of their data. The deeper truth, supported by McKinsey, CBRE, and Cushman & Wakefield research, is that most of the value in CRE comes from execution and relationships, not raw data. A rent roll is not a moat. A T12 is not a moat. The relationships, the deal sourcing engine, and the operating playbooks are. Treating standard CRE financials as crown jewels often blocks adoption without offering any real competitive protection. According to Cushman and Wakefield's analysis, the firms pulling ahead are the ones treating AI as an operating layer across the business, not a one-off experiment.
What This Means for CRE Investors in 2026
The macro picture is unambiguous. The AI in real estate market is projected to reach $1.3 trillion by 2030 at a 33.9% CAGR. 92% of corporate occupiers have already initiated AI programs, but only 5% report achieving most of their AI program goals. CRE sales volume is forecast to increase 15 to 20% in 2026, and the firms that can underwrite, close, and operate faster will capture disproportionate deal flow. The data fears Ben Miller called out are the single biggest barrier to closing that capability gap.
The capital is moving with or without CRE. Anthropic's enterprise revenue has surged dramatically through 2026 as agentic AI shifts from experiment to operating layer. See our coverage of Anthropic's enterprise surge and what it means for CRE investors for the full picture. The investors who treat AI deployment as a 2026 operating priority, not a 2027 maybe, will compound the lead they already hold.
Frequently Asked Questions
Q: Will Anthropic or OpenAI train their models on the data I upload?
A: Not under enterprise plans. Anthropic's Claude for Work, ChatGPT Enterprise, and Gemini for Workspace all contractually commit not to use customer data for model training. Consumer plans have different defaults, which is why enterprise procurement matters for CRE firms.
Q: What does "data-ready" mean in the Keyway survey?
A: Data-ready means your organization's core data (rent rolls, T12s, lease abstracts, deal pipelines, market comps) is consistently formatted, current, and accessible in a way that AI tools can consume without manual cleanup. The Keyway 8% figure shows that almost no CRE firms have reached this bar yet.
Q: What is the fastest way for a CRE firm to move from AI pilot to enterprise deployment?
A: Focus on one high-leverage workflow (underwriting memo drafting, lease abstraction, or OM summarization), measure cycle time and analyst hours saved over 60 to 90 days, and only then expand to additional use cases. Trying to do everything at once is the most common reason pilots stall.
Q: Are the data security concerns about Claude and ChatGPT legitimate or overblown?
A: They are legitimate if you are using consumer-tier accounts, and largely addressable if you are using enterprise-tier plans with proper data-handling controls, vendor diligence, and a documented data-use policy. Treat the question as a procurement and contracting issue, not a binary "AI is safe or unsafe" question.
Q: How fast is the CRE AI adoption gap closing?
A: Slowly at the enterprise-deployment level (9% per the Keyway survey) but quickly at the pilot level (nearly 50%). Industry consensus suggests the gap will narrow materially in 2026 and 2027 as enterprise procurement and data infrastructure catch up to staff-level usage.