$475 million seed round. Company is two months old. No product.

That's Unconventional AI. Naveen Rao, ex-head of AI at Databricks. a16z and Lightspeed gave him $4.5 billion valuation. Same month, Yann LeCun targets $3.5 billion for AMI Labs before writing a single line of commercial code.

Cool story. Now what about the other 99% of AI founders?

The ones sitting on a Zoom call hearing: "Great pitch. Show me your numbers."

$202 billion went into AI in 2025. But most VCs still want to see metrics. I dug through Bessemer, a16z, Pilot, and a dozen other sources to find the exact thresholds. Five metrics. Two levels: "good enough to get a meeting" and "excellent, investors call you first."

1. Monthly revenue growth
Good: 15-20% MoM. Excellent: 25%+.
Lovable went from $1M to $100M ARR in 8 months. With 15 employees. Cursor hit $500M ARR by May 2025. Gamma crossed $50M on less than $25M raised. When a VC just saw those numbers, your "solid 10% MoM" looks pedestrian.

2. Gross margin
Good: 50-60%. Excellent: 70%+.
Traditional SaaS runs 80-90%. AI doesn't. Every inference costs real money. Bessemer splits AI companies into "Shooting Stars" (60% margin, steady growth) and "Supernovas" (25% margin, explosive growth). OpenAI runs at ~50%. Anthropic at ~60%. Investors care less about the current number and more about the trajectory.

3. Net Revenue Retention (NRR)
Good: 110%+. Excellent: 130%+.
Median B2B AI: 106%. Top quartile: 120%+. If NRR is above 100%, your revenue grows even without new customers. Companies with high NRR grow 2.5x faster. At early stage, even a cohort showing customers spend 15-20% more each month tells the story.

4. DAU/MAU (stickiness)
Good: 25%+. Excellent: 40%+.
Average SaaS: 13%. Productivity tools: 40-60%. Cursor sits at the top because developers use it every working day. That's partly why it's a $29B company. If your daily-use tool has 10% DAU/MAU, that's not a pricing problem. It's a product problem.

5. CAC payback
Good: under 12 months. Excellent: under 6 months.
SaaS median: 18 months. Best early-stage: 4.8 months. AI products often have natural virality. Cursor spreads through eng teams. Lovable goes viral on social. If your CAC payback is long despite building AI, investors will wonder about your distribution.

Hit 3 out of 5 and you're in the conversation.
Hit all 5 and investors find you before you find them.

The AI era created the widest gap in startup history between fame-funded and metrics-funded companies. For everyone without $1.7B in exits on their resume, these five numbers are the only pitch deck that matters.

Which one is hardest for your startup right now?

Sources:
https://www.bvp.com/atlas/the-state-of-ai-2025
https://a16z.com/revenue-benchmarks-ai-apps/
https://pilot.com/blog/ai-metrics-fundraising-startups
https://pilot.com/blog/ai-pricing-economics-2025
https://www.tanayj.com/p/the-gross-margin-debate-in-ai