{
  "docId": "019dd923-5e88-73ef-bd5d-0d0f98caffe1",
  "docSlug": "5df6bd1b0447b5f6",
  "documentTitle": "2025 Air Street Capital The State of AI Report 2025",
  "authorId": "AirStreetCapital",
  "authorName": "Air Street Capital",
  "documentKindSlug": "consulting-deck",
  "documentKindLabel": "Consulting deck",
  "sourceTypeSlug": "vc_research",
  "sourceTypeLabel": "VC research",
  "presentationDate": null,
  "orientation": "landscape",
  "aspectRatio": 1.777,
  "pageNumber": 111,
  "pageCount": 313,
  "prevPage": 110,
  "nextPage": 112,
  "slideType": "industry_trends",
  "function": "analyze_data",
  "density": "dense",
  "nDataPoints": 11,
  "notes": "Includes a bar chart of gross margins and a line chart of Replit's revenue growth.",
  "elementsJson": [
    "headline_text",
    "paragraph",
    "bar_chart_horizontal",
    "line_chart",
    "footnote"
  ],
  "metadataConfidence": 1,
  "imagePath": null,
  "slideHref": "/slides/019dd923-5e88-73ef-bd5d-0d0f98caffe1/111",
  "deckHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1",
  "deckJsonHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1.json",
  "deckAnchorHref": "/decks/019dd923-5e88-73ef-bd5d-0d0f98caffe1#slide-111",
  "components": [
    {
      "bbox": null,
      "kind": "callout",
      "text": "The primary levers to improve margins are moving off third-party APIs to owned or fine-tuned models, aggressive caching and retrieval efficiency, and looking to ads or outcomes-based pricing.",
      "attrs": null,
      "subkind": null,
      "toolName": "Visual emphasis",
      "toolSlug": "visual-emphasis",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a826-37c660978429",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.4,
        "w": 0.45,
        "x": 0.02,
        "y": 0.45
      },
      "kind": "chart",
      "text": "Marginal Differences: AI startups' recent gross margins",
      "attrs": null,
      "subkind": "bar-horizontal",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "4b2c75ef-4961-43b3-8a84-52965fe1bd2b",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.4,
        "w": 0.45,
        "x": 0.52,
        "y": 0.45
      },
      "kind": "chart",
      "text": "Replit's Revenues: Annualized revenue at the startup spiked this year after it launched an AI coding agent.",
      "attrs": null,
      "subkind": "line",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "d7d61435-2835-436e-80ac-5c4fdf71c9cd",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": null,
      "kind": "metric",
      "text": "Gross Margin: 60%",
      "attrs": null,
      "subkind": "primary",
      "toolName": "Quantification",
      "toolSlug": "quantification",
      "confidence": null,
      "componentId": "019dd952-41ac-7180-a826-3bfaa57f8606",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.18,
        "w": 0.96,
        "x": 0.02,
        "y": 0.22
      },
      "kind": "paragraph",
      "text": "Gross margins are generally commanded by the underlying model API and inference costs and strained by token-heavy usage and traffic acquisition. Surprisingly, several major AI companies don't include the costs of running their service for non-paying users when reporting their GM. Coding agents are under pressure even when revenue grows quickly. The primary levers to improve margins are moving off third-party APIs to owned or fine-tuned models, aggressive caching and retrieval efficiency, and looking to ads or outcomes-based pricing.",
      "attrs": null,
      "subkind": "paragraph",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "6c921b9e-f693-46dd-b48d-0c13bc31c14d",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.08,
        "w": 0.45,
        "x": 0.02,
        "y": 0.85
      },
      "kind": "source-note",
      "text": "Note: Perplexity, Replit, Lovable and StackBlitz do not incorporate the costs of running AI models for nonpaying users in calculations, while OpenAI does. Anthropic's accounting couldn't be learned. Source: The Information reporting",
      "attrs": null,
      "subkind": null,
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "a9f70169-669b-4ee7-bbeb-a81ced8838e2",
      "frameworkName": null,
      "frameworkSlug": null
    },
    {
      "bbox": {
        "h": 0.04,
        "w": 0.4,
        "x": 0.02,
        "y": 0.14
      },
      "kind": "title",
      "text": "The M word: so what about the margins?",
      "attrs": null,
      "subkind": "headline",
      "toolName": null,
      "toolSlug": null,
      "confidence": null,
      "componentId": "8cbeb063-f971-40d1-ab98-5eff211bf779",
      "frameworkName": null,
      "frameworkSlug": null
    }
  ],
  "metrics": [],
  "tools": [
    {
      "name": "2x2 matrix",
      "slug": "matrix-2x2",
      "agent": null,
      "layer": "slide",
      "matchId": "ba7b0ac4-2a0d-4d83-9161-6160b830f3aa",
      "evidence": "The slide presents a comparison of gross margins for different AI startups, potentially indicating a 2x2 matrix, although not explicitly shown.",
      "confidence": 0.5
    }
  ],
  "frameworks": [],
  "arcBeats": [
    {
      "to": 189,
      "from": 12,
      "beatId": "019dd95a-0682-776c-8e35-9f380673831f",
      "arcName": "The Triple Take",
      "arcSlug": "triple-take",
      "beatName": "The Facts (What)",
      "beatSlug": "triple-take-the-facts-what",
      "evidence": "Sections 1-2 lay out research findings and industry data with charts and case studies.",
      "position": 1,
      "confidence": 70,
      "parentBeatName": "Setup",
      "parentBeatSlug": "setup"
    },
    {
      "to": 189,
      "from": 90,
      "beatId": "019dd95a-0682-776c-8e35-b06421e14afb",
      "arcName": "The Onion",
      "arcSlug": "onion",
      "beatName": "Deeper Layer",
      "beatSlug": "onion-deeper-layer",
      "evidence": "Industry section exposes economics, infra, geopolitics beneath research.",
      "position": 3,
      "confidence": 45,
      "parentBeatName": "Development",
      "parentBeatSlug": "development"
    }
  ],
  "loops": [
    {
      "to": 112,
      "from": 105,
      "name": "Funnel Analysis",
      "slug": "32-funnel-analysis",
      "bestFor": "Sales optimization, customer journey, process efficiency",
      "matchId": "019dd95a-07fe-70ce-8d3e-a3f81a44ee54",
      "evidence": "GPT-5 economics → vibe coding costs → Cursor unit economics → 'When do we see profitable models?' (p.112).",
      "position": 6,
      "objective": "Trace AI economics from revenue through margins to profitability",
      "structure": "Top of Funnel -> Stage 1 Conversion -> Stage 2 Conversion -> Bottom of Funnel -> The Leak",
      "confidence": 65,
      "description": "Show progressive filtering through stages to identify where drop-off occurs"
    }
  ],
  "imagePathAlt": null,
  "thumbSrc": null,
  "thumbSrcAlt": null,
  "locked": true
}