a16Z recently surveyed over 100 leading CIOs across 15+ industries to get the latest pulse on enterprise AI spend in software. The learnings aren’t a surprise — but they are useful to see just how core AI spend has become in the enterprise.
It’s not just part of the innovation budget anymore.
Top 5 SaaStr Learnings: Enterprise AI in 2025
1. AI Budgets Are Exploding—And Moving to Core IT Spend
The Hard Data: Enterprise AI spend is growing 75% year-over-year, with innovation budget allocation dropping from 25% to just 7% of total AI spend. One CIO reported: “what I spent in 2023 I now spend in a week.”
The Money Shift: AI has officially graduated from pilot programs to permanent budget line items in core IT and business units. This represents a fundamental shift from experimental dollars to recurring revenue opportunity.
The B2B Reality: You’re no longer selling to innovation teams with small budgets—you’re competing for the same enterprise IT dollars as traditional software, which means enterprise sales cycles, security requirements, and procurement rigor are now table stakes.
2. Multi-Model Deployment Is Crushing Single-Vendor Strategies
The Hard Data: 37% of enterprises now use 5+ models in production (up from 29% last year). OpenAI maintains overall market leadership, but the breakdown is telling: 67% of OpenAI users deploy non-frontier models vs. only 41% for Google and 27% for Anthropic.
The Price War: Google’s Gemini 2.5 Flash costs $0.26/million tokens vs. GPT-4.1 mini at $0.70/million tokens—a 63% cost advantage that’s driving enterprise adoption.
The B2B Reality: Enterprise customers expect you to intelligently route between models based on use case and cost. Your differentiation isn’t which model you use—it’s how smartly you orchestrate multiple models to optimize performance and cost.
3. “Buy vs. Build” Has Completely Flipped in 12 Months
The Hard Data: Over 90% of enterprises are now testing third-party AI apps for customer support rather than building internally. One public fintech abandoned their internal build mid-development to purchase a third-party solution instead.
The Switching Cost Reality: As AI workflows become more agentic and complex, switching costs are rising dramatically. One leader noted: “all the prompts have been tuned for OpenAI…changing models is now a task that can take a lot of engineering time.”
The B2B Opportunity: Enterprises recognize they can’t keep up with AI optimization internally. Focus on deep vertical solutions where continuous model optimization and prompt engineering create defensible moats.
4. AI-Native Companies Are Crushing Incumbents by 2-3x Speed
The Hard Data: Leading AI-native companies are reaching $100M ARR significantly faster than previous software generations, driven by prosumer adoption creating enterprise pull-through demand.
The Quality Gap: Users who adopted AI-native coding tools like Cursor show “notably lower satisfaction” with previous-generation tools like GitHub Copilot, indicating a fundamental step-function improvement in capabilities.
The B2B Takeaway: This is the classic platform shift moment—incumbents retrofitting AI are losing to companies built AI-first from the ground up. Lead with demonstrably superior outcomes, not feature parity.
5. Enterprise AI Procurement Now Mirrors Traditional Software Buying
The Hard Data: External benchmarks like LM Arena have become the new “Magic Quadrant” for initial model filtering, while security and cost have gained ground on accuracy as key purchasing criteria. Usage-based pricing still dominates with CIOs uncomfortable with outcome-based models.
The Trust Shift: More enterprises are now hosting directly with model providers (OpenAI, Anthropic) rather than going through cloud providers—a complete reversal from last year when enterprises preferred cloud provider intermediaries for trust and compliance.
The Evaluation Reality: While internal benchmarks and golden datasets remain critical, enterprises increasingly reference external evaluations as their first filter, similar to how they use Gartner for traditional software purchases.
The B2B Implication: AI procurement has matured to enterprise software standards. You need rigorous evaluation frameworks, security checklists, benchmark performance data, and clear usage-based pricing models. The “move fast and break things” era is over—enterprises want traditional software buyer protection with AI innovation speed.
