If you're a CTO, CFO, or CIO evaluating AI platforms right now, you're probably choosing between ChatGPT and Claude. You might even feel a little defensive about your preference.
Turns out, so do your peers. And for the first time, we have enterprise spending data that shows exactly who's winning—and why it's not about the model.
Ramp's March 2026 AI Index dropped this week with a headline number that's reshaping vendor conversations: Anthropic now wins roughly 70% of head-to-head matchups against OpenAI among businesses buying AI for the first time.
That's a complete reversal from 2025.
The Numbers That Matter
Here's what changed in enterprise AI purchasing behavior:
Anthropic's enterprise adoption surge:
- Nearly 1 in 4 businesses on Ramp now pay for Anthropic (up from 1 in 25 a year ago)
- 70% win rate in head-to-head matchups vs OpenAI for new enterprise customers
- Growth accelerating despite higher pricing and active rate limits
OpenAI's first major decline:
- Adoption rate fell 1.5% — its largest single-month decline ever
- Still dominant in consumer/individual markets
- Losing ground specifically in first-time enterprise buyers
What makes this data meaningful: It's not survey responses or analyst predictions. This is actual enterprise spending behavior from thousands of businesses managing AI vendor payments through Ramp.
Photo by Tima Miroshnichenko on Pexels
Why This Matters (It's Not the Benchmarks)
Here's the part that stopped me: Anthropic is winning despite charging more, with a comparable product, while actively turning away revenue through rate caps.
In most enterprise software categories, the cheaper product wins. Not here.
What's different:
-
User Experience Over Performance
- Claude Code and OpenAI's Codex are roughly comparable products
- Codex is arguably cheaper (Altman recently reset usage limits)
- Yet businesses are choosing Claude at 70% rate
- The deciding factor: Daily workflow integration, not raw capabilities
-
The Platform Lock-In Effect
- Both platforms now have 200+ apps in their connector ecosystems
- Only 11% overlap between ChatGPT and Claude ecosystems
- ChatGPT: Super-app approach (shopping, travel, food)
- Claude: Developer tools and financial data terminals
- Once configured, switching costs compound fast
-
Cultural Differentiation
- Ramp economist Ara Kharazian thinks the moat is cultural
- The DoD backlash created a "certain class of user" who noticed
- Choosing platforms may become identity signals (like iMessage blue vs green bubbles)
- Enterprise buyers care about vendor values, not just features
Photo by Tima Miroshnichenko on Pexels
The Real Product Is the Harness, Not the Model
The model stopped being the bottleneck sometime last year. The system you build around it—what the industry now calls a "harness"—is the real product.
What enterprises are actually buying:
- Skills and connectors: Custom workflows integrated with internal tools
- Project files: Context that teaches the AI how your specific work gets done
- Memory systems: Accumulated organizational knowledge
- Ecosystem integrations: Third-party apps and data sources
Every skill you install, every tool you connect, every memory the system builds—that's infrastructure lock-in that's expensive cognitively to unload.
As a16z partner Olivia Moore put it: "Context and memory compound."
For business leaders, this means:
- Vendor selection is a multi-year commitment, not a quarterly evaluation
- Total Cost of Ownership includes switching costs (retraining, reconfiguring, lost context)
- "Try both and see which is better" is the wrong framework—once you commit, migration is painful
What Changed This Week
Anthropic doubled down on its UX advantage with inline interactive visuals, letting Claude build charts, diagrams, and interactive visualizations right inside your conversation.
It sounds like a small feature. But combined with skills, connectors, and plugins, it makes Claude less of a text box and more of a workspace where information is both processed and presented.
Meanwhile, ChatGPT is pushing toward a super-app strategy with shopping, travel, and entertainment integrations—serving consumer use cases more than enterprise workflows.
Two platforms. Two philosophies. Two ecosystems.
Photo by Tima Miroshnichenko on Pexels
Decision Framework for Enterprise Buyers
If you're evaluating AI platforms right now, here's what actually matters:
For Technical Leaders (CTO, VP Engineering, Head of AI)
Evaluate on integration, not performance:
- Which platform connects to your internal tools? (Slack, Jira, GitHub, CRM)
- What's your team's daily workflow? (Developer tools vs business workflows)
- How do you govern AI usage across departments? (Rate limits, cost controls, audit trails)
Ask about the harness, not the model:
- How customizable are project files and system prompts?
- Can you isolate AI context per department or project?
- What's the migration path if you need to switch vendors in 2 years?
For Business Leaders (CFO, COO, CRO)
Total Cost of Ownership (TCO) beyond per-token pricing:
- What's the cost of configuring and maintaining the platform?
- What productivity gains come from ecosystem integrations?
- What's the switching cost if the vendor relationship sours?
Strategic vendor risk:
- Is the vendor's business model aligned with long-term enterprise adoption?
- Are they financially stable enough to support multi-year commitments?
- Do their values align with your company's risk tolerance? (DoD work, data privacy, etc.)
For Department Leaders (CMO, VP Sales, VP Ops)
Workflow integration over features:
- Does the platform integrate with your daily tools? (HubSpot, Salesforce, Google Workspace)
- Can non-technical users configure it without IT support?
- What's the learning curve for your team?
The Bigger Picture: What This Data Tells Us
Three strategic insights from Ramp's AI Index:
-
Enterprise AI vendor preference is diverging from consumer preferences
- ChatGPT still dominates consumer/individual users
- Claude winning first-time enterprise buyers at 70% rate
- Different buying criteria: UX + ecosystem vs raw performance
-
Switching costs are real and underestimated
- Once you configure skills, connectors, and memory systems, migration is painful
- Vendor lock-in happens faster than traditional enterprise software
- Treat AI platform selection like choosing an ERP—multi-year commitment
-
Cultural values are becoming a moat
- The DoD backlash differentiated Anthropic's brand among "a certain class of user"
- Enterprise buyers care about vendor ethics and transparency
- In a commoditized model market, identity signals matter
What to Do About This
If you're currently evaluating AI platforms:
- Don't optimize for the best demo—optimize for long-term workflow integration
- Test both platforms with your actual internal tools and data (not generic benchmarks)
- Ask about switching costs upfront: What happens if we need to migrate in 2 years?
If you've already chosen a platform:
- Double down on customization (skills, connectors, project files)
- Audit your switching costs quarterly (how painful would migration be?)
- Stay informed on competitive moves—platforms are diverging fast
If you're planning multi-year AI strategy:
- Assume vendor lock-in will compound (plan for it, don't fight it)
- Choose vendors whose values align with your company's risk tolerance
- Invest in governance infrastructure that works across platforms (for when you need to support multiple)
Want to calculate your own AI ROI? Try our AI ROI Calculator — takes 60 seconds and shows projected savings, payback period, and 3-year ROI.
Related: 70% of Enterprise AI Buyers Can't Measure ROI: Survey Data
Continue Reading
- GPT-5.4 vs Claude Opus 4.6: 2026 Benchmark Comparison
- GPT-5.4 vs Claude Opus 4.6: Enterprise Buyer Guide 2026
- Claude Cowork Review 2026: Desktop AI for Enterprises
Photo by Tima Miroshnichenko on Pexels
Photo by Tima Miroshnichenko on Pexels
Photo by Tima Miroshnichenko on Pexels