gotnerfed

Subsidy Risk Index

Which AI vendors are most likely to silently nerf you next? Predictive risk score based on wrapper status, funding stage, observed cadence, and parent-company depth.

The Twitter/X discourse keeps surfacing this: Replit CEO admitted “we were paying more to Anthropic/OpenAI than we made.” Internal Cursor estimates put $200/mo subs at $2,000+ in compute. VC subsidies are ending. Every AI wrapper SaaS is a ticking nerf bomb — this index ranks them.

Methodology: pure deterministic rubric. Read the math. Source code: lib/subsidy-risk.ts.

Lovable
Lovable · Seed
65
margin stressed
  • ·Wraps a foundation model (compute pass-through risk)
  • ·Early-stage (Seed) with VC runway pressure
Cursor
Anysphere · Series B
56
margin stressed
  • ·Wraps a foundation model (compute pass-through risk)
  • ·Mid-stage (Series B); growth required to raise next round
  • ·1 prior critical receipt — repeat risk elevated
Claude Code
Anthropic · part of Anthropic
55
margin stressed
  • ·Wraps a foundation model (compute pass-through risk)
  • ·1 receipt in last 90 days — monitoring
  • ·2 prior critical receipts — repeat-offender pattern
Replit Agent
Replit · Series B
48
on watch
  • ·Wraps a foundation model (compute pass-through risk)
  • ·Mid-stage (Series B); growth required to raise next round
Bolt.new
StackBlitz · part of StackBlitz · Series B
48
on watch
  • ·Wraps a foundation model (compute pass-through risk)
  • ·Mid-stage (Series B); growth required to raise next round
Perplexity
Perplexity · Series C
42
on watch
  • ·Wraps a foundation model (compute pass-through risk)
Windsurf
Codeium · Series C
42
on watch
  • ·Wraps a foundation model (compute pass-through risk)
Runway
Runway · Series D
38
on watch
  • ·Wraps a foundation model (compute pass-through risk)
GitHub Copilot
GitHub · part of Microsoft
32
on watch
  • ·Wraps a foundation model (compute pass-through risk)
  • ·1 receipt in last 90 days — monitoring
  • ·Owned by Microsoft — deeper pockets, lower acute pressure
v0
Vercel · part of Vercel
30
on watch
  • ·Wraps a foundation model (compute pass-through risk)
ChatGPT
OpenAI · private
23
stable
  • ·Foundation model provider (lower margin pressure)
  • ·1 receipt in last 90 days — monitoring
Claude (Anthropic)
Anthropic · Series F
18
stable
  • ·Foundation model provider (lower margin pressure)
  • ·1 receipt in last 90 days — monitoring
Suno
Suno · Series B
18
stable
  • ·Mid-stage (Series B); growth required to raise next round
Notion AI
Notion · part of Notion Labs · Series C
17
stable
  • ·Foundation model provider (lower margin pressure)
OpenAI API
OpenAI · private
13
stable
  • ·Foundation model provider (lower margin pressure)
ElevenLabs
ElevenLabs · Series C
12
stable
    Grok (xAI)
    xAI · part of X Corp · Series C
    9
    stable
    • ·Foundation model provider (lower margin pressure)
    • ·Owned by X Corp — deeper pockets, lower acute pressure
    Midjourney
    Midjourney · private (no VC)
    5
    stable
      DeepSeek
      DeepSeek
      5
      stable
      • ·Foundation model provider (lower margin pressure)
      Gemini (Google)
      Google · part of Alphabet
      0
      stable
      • ·Foundation model provider (lower margin pressure)
      • ·Owned by Alphabet — deeper pockets, lower acute pressure

      The contrarian read

      Ranking AI vendors by Nerf Index tells you who has nerfed users in the past. Subsidy Risk tells you who is most likely to nerf you next. The wrappers carrying the highest scores aren't bad companies — they're companies whose unit economics force them to silently degrade as VC subsidies expire. We're predicting the bait-and-switch before it happens.