AI Threat Assessment · 26 May 2026

Perplexity AI

AI Search Engine
ALREADY A ZOMBIE
8.7/ 10

Perplexity built the perfect search engine for the post-Google era — real citations, no ads, conversational answers that actually understand context. They raised $74M in Series B funding and convinced investors that 'Google but with ChatGPT' was a $1B thesis. The problem is they're selling premium subscriptions to a product that OpenAI, Anthropic, and Google are all shipping as a free feature inside the thing people already have open.

Business Model
9.5
Automation Risk
8.0
Moat Strength
2.0
Adaptability
7.5
Need Survival
9.0
AI Threat Level
BUSINESS MODEL REPLACEABILITY

Their Pro plan charges $20/month for 'unlimited AI searches' and 'advanced reasoning' — features that ChatGPT Plus and Claude Pro bundle as throwaway capabilities alongside actual frontier model access. Charging for search when the search is a commodity wrapper around someone else's LLM is like charging for calculator access on a smartphone.

9.5
WORKFORCE AUTOMATION RISK

Their engineering team builds search interfaces and citation parsing — work that gets absorbed when OpenAI's next update includes web search by default, or when Google's Gemini finally learns to cite sources without hallucinating.

8.0
MOAT STRENGTH

They have genuinely superior citation methodology and cleaner UX than Google's AI Overviews — but these are product features, not moats. The moment any foundation model provider decides search is table stakes, Perplexity becomes a bookmark folder.

2.0
AI ADAPTABILITY SIGNALS

They're hiring for 'AI research scientists' and 'LLM infrastructure' roles, which translates to 'we're trying to build our own foundation model to avoid getting cut off by the people whose APIs we currently resell.' The funding lets them try; the physics doesn't let them succeed.

7.5
WILL THE NEED SURVIVE AI?

Search survives — as a primitive inside every AI assistant. Perplexity built a beautiful, dedicated search experience for the six months between 'Google search is broken' and 'every AI can search.' They optimized for a window that's already closing.

9.0
Verdict

Platform absorption kills them within 18 months — not through competition, but through ChatGPT's next update that makes opening a separate search app feel like launching Internet Explorer to check email. They built the perfect search engine for an era that ended the moment OpenAI decided search was a feature, not a product.

Scores are based on public information and AI analysis. This is an affectionate roast, not a financial assessment. The best companies use this as a mirror, not a verdict.

Roast another →