What is enterprise AI adoption, and why does it matter that Claude just passed ChatGPT? Enterprise AI adoption is the share of businesses that actually pay for and use a given AI model in their operations, and in 2026 it became a real contest. According to the Ramp AI Index, which measures live corporate card and invoice spend across more than 50,000 US companies, Anthropic's Claude edged past OpenAI's ChatGPT in US business adoption for the first time in its May 2026 reading, then extended the lead in June. For commercial real estate firms choosing which AI to standardize on for underwriting, lease abstraction, and due diligence, the question of Claude vs ChatGPT enterprise adoption is no longer abstract. It is a procurement decision, and it is the heart of any serious AI model comparison for CRE investors.
Key Takeaways
- Anthropic's Claude passed OpenAI's ChatGPT in US business adoption for the first time in the Ramp AI Index, which tracks real spending rather than survey responses.
- In the May 2026 reading Claude reached about 34% of businesses versus ChatGPT near 32%, and by June Anthropic led near 41% with OpenAI close behind around 40%.
- The shift is dramatic over time: a year earlier Anthropic sat near 8% of businesses while OpenAI led near 32%, so Claude roughly quadrupled its share.
- Adoption is broad but shallow; an IDC survey found only about 19% of organizations report extensive Claude use, echoing the wider CRE AI productivity gap.
- With 52% of customers using both vendors and low switching costs, CRE firms should stay multi-model rather than lock into a single AI provider.
- Anthropic's lead is fragile, threatened by rising costs and compute limits, and OpenAI has launched a $4 billion effort to win enterprises back.
Claude vs ChatGPT: What the Ramp Data Shows
The Ramp AI Index is useful precisely because it is not a survey. It draws on actual card and invoice data from more than 50,000 US companies, so it reflects what businesses pay for, not what they say. Its May 2026 reading showed Anthropic at roughly 34% of business adoption against OpenAI near 32%, the first crossover on record, with overall AI adoption among tracked businesses passing 50% for the first time. The June 2026 reading kept Anthropic in front near 41% of firms with paid AI subscriptions, with OpenAI close behind around 40%, a lead of only about 1.5 percentage points. Ramp also found Anthropic increasingly converting first-time AI buyers directly to Claude, rather than only poaching OpenAI users.
Context matters. A year earlier, the same index had OpenAI well ahead near 32% with Anthropic a distant 8%, so this is a fast reversal, not a long-running trend. It also is not winner-take-all: Ramp reports that 52% of customers using either vendor use both, and that 43% of Anthropic's customers switched from another AI vendor. In a market with low switching costs and rapid model releases, today's leaderboard is a snapshot, not a verdict. We first flagged this turn in our earlier coverage of how Claude began overtaking ChatGPT in enterprise AI; the new spending data confirms it.
Why Claude Is Winning the Enterprise
The engine is a single product: Claude Code, Anthropic's agentic coding tool, which the company calls the fastest-growing product in its history. One analysis estimated that roughly 4% of all public GitHub commits worldwide were being authored by Claude Code, double the share of a month earlier. Beyond coding, enterprises cite Claude's reliability, long-context handling, and tendency to follow instructions faithfully as advantages for production work where a wrong answer is costly. For CRE, those same traits matter when a model abstracts hundreds of leases or reconciles a rent roll, tasks where consistency beats flash. The current flagship, Claude Opus 4.8, is priced near $5 per million input tokens and $25 per million output tokens, while OpenAI counters with GPT-5.5 and its Codex line. Capability gaps between top models are narrow and temporary, which is exactly why adoption, not benchmarks, is the more telling signal.
What the Adoption Flip Means for CRE Investors
The practical lesson is not pick Claude. It is do not lock in. Because 52% of business users already run both Claude and ChatGPT, and because switching costs are low, the smart posture for a CRE firm is multi-model: route each task to whichever model does it best and cheapest. A long-context lease abstraction or a careful underwriting review may favor Claude today; a quick marketing draft or image task may favor GPT-5.5 or Gemini. This is the same model-agnostic approach we described when Apple opened the iPhone to multiple AI providers in our coverage of WWDC 2026 AI Extensions, now applied to the enterprise stack.
There is a sober caveat under the adoption headline. Broad adoption is not the same as deep use. An IDC survey found only about 19% of organizations report extensive Claude use, with another 25% merely evaluating it. In other words, many firms pay for AI without operationalizing it, which is the precise gap we documented in the AI productivity gap and CRE AI ROI. Standardizing on the most-adopted model means little if your team only uses it as a chat box. The AI Consulting Network works with CRE firms to close that gap, turning a license into a measurable workflow.
A CRE AI Vendor Playbook for 2026
- Match the model to the task: Reserve a frontier model like Claude Opus 4.8 for high-stakes underwriting and due diligence, and route bulk extraction to cheaper models to control spend.
- Stay multi-model: Keep at least two providers active so you can shift work as prices and capabilities change, and so no single outage stalls your pipeline.
- Govern access: Decide who needs frontier-model seats, set spend caps on automated workflows, and document how AI outputs are reviewed before they touch a deal.
- Measure outcomes: Track cost per completed task against the analyst hours and error risk removed, not vanity usage.
- Revisit quarterly: With leadership changing every few months, treat your AI vendor choice as a decision you re-test, not one you make once.
For market context on how leading real estate firms operationalize AI, research hubs such as CBRE publish regular insight, and the underlying spending data is tracked publicly by Ramp.
The Lead Is Fragile
None of this is settled. The same Ramp analysis that crowned Anthropic warns that its position may be more fragile than it looks, pressured by escalating costs, compute constraints, and the very token-based pricing that fueled its rise, the same pricing shift CRE buyers are now budgeting around. OpenAI has not stood still: it launched a $4 billion deployment unit with 19 private equity and consulting partners to embed engineers inside enterprises and accelerate Codex adoption, with its revenue chief calling the market the most competitive he has seen. For CRE investors, the takeaway is steady: adoption leadership will trade hands, so build a flexible AI stack rather than a bet on one winner. CRE leaders who want help designing that stack can reach out to Avi Hacker, J.D. at The AI Consulting Network.
Frequently Asked Questions
Q: Did Claude really pass ChatGPT in business adoption?
A: Yes, by the Ramp AI Index, which tracks real corporate spending across more than 50,000 US companies. Claude edged past ChatGPT for the first time in the May 2026 reading, near 34% to 32%, and by June led near 41% with OpenAI close behind around 40%.
Q: Should a CRE firm switch entirely to Claude?
A: Not necessarily. With 52% of business users running both Claude and ChatGPT and switching costs low, a multi-model approach usually wins. Route each task, from lease abstraction to marketing, to whichever model performs best and costs least, and keep a second provider active.
Q: Why is Claude gaining so fast with businesses?
A: Much of the growth traces to Claude Code, Anthropic's agentic coding tool, plus a reputation for reliability and strong long-context performance on production tasks. Enterprises increasingly choose Claude for new builds even when they keep using OpenAI tools elsewhere.
Q: Does high adoption mean firms are getting value from AI?
A: Not automatically. An IDC survey found only about 19% of organizations report extensive Claude use, so many pay for AI without operationalizing it. The value comes from wiring the tool into real workflows and measuring results, not from the license alone.