P/S Ratio
Valuation divided by annual revenue · the single number that captures whether an AI company is fairly priced, frothy, or bubbled.
Valuation divided by annual revenue · the single number that captures whether an AI company is fairly priced, frothy, or bubbled.
Basic
A healthy SaaS company trades at 10-20× P/S. AI companies trade at 30-200+× P/S. OpenAI at $500B valuation / $15B ARR = 33× P/S. Anthropic at $180B / $8B ARR = 22× P/S. Cursor at $100B / $500M ARR = 200× P/S. The BenchGecko AI Bubble Index uses aggregate P/S as its 40% weight component.
Deep
P/S ratio benchmarks: healthy SaaS 5-15×, hypergrowth SaaS 20-50×, frothy 50-100×, bubble 100+×. AI sector aggregate P/S in 2025-2026 hit 60-80× · nearly 5× above healthy SaaS benchmark. Drivers: hyperscaler AI revenue counted as "software" growth, investor FOMO, and genuine belief in 10× productivity impact. Outlier cases (Cursor 200×, Perplexity 100×) reflect pricing power premium over incumbent AI.
Expert
P/S ratio ignores profitability · a flawed metric for mature companies but useful for hypergrowth. AI companies typically lose money; P/E is undefined, so P/S is the default. Comparison to dotcom: Cisco 1999 peaked at ~30× P/S, crashed to 5×. AI sector today at 60-80× aggregate suggests either (a) genuine productivity shift justifies premium or (b) mean-reversion looms. The BenchGecko Bubble Index weights this at 40% because of its historical reliability.
AI sector P/S sits at 4-5× healthy SaaS benchmarks. Every valuation story hinges on this ratio normalizing or sustaining.
Depending on why you're here
- ·P/S = valuation / annual revenue
- ·Healthy SaaS 10-20×, hypergrowth 20-50×, bubble 100+×
- ·Default metric when companies are unprofitable
- ·P/S tells you which AI companies are overpriced · matters for procurement
- ·High P/S companies have pricing power to exploit · matters for vendor lock-in risk
- ·Watch P/S decay as competition commoditizes
- ·AI sector aggregate 60-80× vs SaaS 15× baseline · thesis-defining
- ·Individual P/S hierarchy: Cursor > Perplexity > OpenAI > Anthropic > Google AI
- ·1999 Cisco parallel · 30× peak, 5× floor · possible AI trajectory
- ·How much investors are paying for each dollar of revenue
- ·AI companies are priced like they'll grow 10× · that may or may not happen
- ·The number that tells you if AI is a bubble
P/S is the single most important AI economy metric. Every other valuation debate is downstream of this number.
Knowing P/S ratios tells you when an AI vendor is "too valued to fail" (sustainable pricing) vs "must grow 10× quickly" (acquisition risk).