Tomasz Tunguz (Theory Ventures) · 2025-12-30 · 156d

Gross Profit per Token: A New Valuation Metric for AI Inference Companies

Tomasz Tunguz analyzes Meta's $2.5B acquisition of Manus through the lens of gross profit per token, arguing this metric better predicts AI company valuations than raw token volume. By comparing six AI inference companies, he demonstrates that gross profit per token correlates 0.70 with valuation, suggesting investors prioritize monetization efficiency over scale.

20 metrics· Cited 0× in the knowledge base ·Open source ↗

Metrics in this report

Correlation: GP per Token vs Valuation

0.70correlation coefficient

r

Six AI inference companies

Correlation: Token Volume vs Valuation

0.47correlation coefficient

r

Raw token volume (lower explanatory power)

GP Multiple - Anthropic

67x

calculated

Gross profit multiple

GP Multiple - DeepSeek

20x

calculated

Gross profit multiple

GP Multiple - Groq

102x

calculated

Gross profit multiple

GP Multiple - Manus

50x

calculated

Gross profit multiple

GP Multiple - Perplexity

222x

calculated

Gross profit multiple (application layer)

GP Multiple - Together AI

24x

calculated

Gross profit multiple

Gross Margin - Anthropic

55%

blended

Direct sales and cloud resale

Gross Margin - DeepSeek

85%

claimed

AI inference via architectural efficiency

Gross Margin - Groq

40%

estimated

LPU-based inference

Gross Margin - Manus

50%

estimated

Agent-based SaaS model

Gross Margin - Perplexity

60%

reported

Application-layer search

Gross Margin - Together AI

45%

estimated

GPU-based inference reseller

Gross Margin - Traditional SaaS

71-72%

range

Publicly traded software companies

Manus ARR

100$M

actual

Achieved in 8 months

Manus Cumulative Tokens

147T

actual

Since launch March 6, 2025

Manus Monthly Tokens

16.3T

estimated

December 2025

Manus Valuation

2.5$B

acquisition price

Meta acquisition

R-squared: GP per Token vs Valuation

48.5%

Variance explained