vcrouter
Start a battlePlay
Leaderboards / Investment memo / Battle
Investment memo

Perplexity — AI answer engine challenging Google, aggressive consumer + enterprise push

September 13, 2026 · $0.100 for both answers and the judge

DeepSeek V4 Pro wins

On September 13, 2026, DeepSeek V4 Pro and Grok 4.6 each wrote an investment memo for the same brief, without seeing the other's answer. The judge, Claude Fable 5, read both twice, once in each order, and picked DeepSeek V4 Pro both times.

Why, in the judge's words“A cites concrete verifiable traction facts (10M MAU, ~$20M ARR, $520M→$3B valuations), names copyright risk, and matches B's rigor and honesty otherwise.”

The two answers

DeepSeek V4 ProWinner
$0.012 · 55s · 3,086 tokens

Investment Memo: Perplexity

Thesis

Perplexity is a rare attempt to unbundle Google’s core product with a genuinely better user experience for a large and growing class of queries. If it can convert even a small share of global search volume into high-margin subscription and enterprise revenue, it becomes a $10B+ revenue company. At a hypothetical Series A, the entry price is low enough that the asymmetry is attractive: the probability of failure is high, but the outcome if Perplexity becomes a default answer engine is fund-returning. This is not a search feature; it is a wedge into the $200B+ search market and the enterprise knowledge market.

Product & Wedge

Perplexity’s product is an AI-native answer engine: it synthesizes real-time web results into cited, conversational answers instead of returning a list of blue links. The wedge is informational and research-heavy queries where users want synthesis, not navigation. Early adopters are students, developers, researchers, and knowledge workers who value speed and source transparency.

The product has moved quickly: conversational follow-up, file upload, image generation, mobile apps, and an API. The enterprise push is real — Perplexity Enterprise Pro offers SSO, SOC2, team billing, and internal knowledge search. That matters because enterprise search is a fragmented, high-pain category where Google and Microsoft have underdelivered. The consumer product builds brand and query data; the enterprise product builds revenue and stickiness.

Market & Competition

Global search advertising is over $200B annually. Enterprise knowledge management and search is a smaller but fast-growing market, often estimated at $10B+. The incumbents are formidable:

  • Google — dominant distribution via Chrome, Android, and default deals; integrating Gemini into Search.
  • Microsoft — Copilot in Bing, Windows, Edge, and Office; owns enterprise distribution.
  • OpenAI — ChatGPT with browsing is the default AI assistant for many; massive consumer mindshare.
  • You.com, Arc Search, Brave Search — smaller AI search challengers.
  • Enterprise: Glean, Microsoft 365 Copilot, Google Vertex AI Search, Elastic, Coveo.

Perplexity’s differentiation is focus: it is not a general chatbot, not an OS feature, not a legacy search engine bolting on AI. But focus alone does not overcome distribution.

Traction & Business Signal

Publicly known figures are limited but directional. Perplexity reported 10 million monthly active users in January 2024 and said it served 500 million queries in 2023. Revenue was reported around $20M ARR in early 2024. Valuation moved from $520M in January 2024 to a reported $3B in April 2024, reflecting strong investor demand. Current MAU, retention, cohort behavior, and enterprise revenue are unknown from public data.

The signal is real: rapid query growth, strong brand among early adopters, and an enterprise product shipping quickly. But the business signal is incomplete — we do not know if usage is durable or novelty-driven.

Risks

Three risks could kill this deal:

  1. Distribution/default moat. Google and Microsoft can bundle AI answers into Chrome, Android, Windows, Edge, and Office. If Perplexity cannot become a default habit or secure distribution, it becomes a feature, not a company. This is the existential risk.
  2. Monetization unit economics. Search ads monetize commercial intent. Perplexity’s answer engine may attract informational queries with lower click-through and advertiser value. Subscription alone may cap TAM. If revenue per query is a fraction of Google’s, the model may not scale to justify a large outcome.
  3. Copyright and platform dependency. Publishers have already raised scraping and licensing concerns. Legal exposure could force costly deals or degrade the product. Perplexity also depends on underlying LLM APIs and search indexes from companies that compete with it. Suppliers can change terms or cut off access.

Verdict

Invest at a hypothetical Series A, conditional on one answer.

The one question I would need answered first: Can you show cohort-level data proving Perplexity is becoming a default search habit — specifically 6-month retention and share of total search queries per user — rather than a novelty overlay?

If the data shows users shifting meaningful query share to Perplexity and returning over months, the distribution risk is manageable and the upside is fund-returning. If not, it is a feature waiting to be bundled by incumbents, and I would pass.

Grok 4.6
$0.009 · 34s · 1,486 tokens

Investment Memo: Perplexity (Hypothetical Series A)

Thesis. Perplexity could be a fund-returner if it becomes the default “answer layer” between users and the open web/LLMs—capturing high-intent queries that Google still monetizes poorly and that ChatGPT treats as conversation rather than retrieval. Search is a $200B+ advertising market with a consumer habit that has not been fully disrupted; an AI-native interface that cites sources, stays current, and feels faster than ten blue links can take share without needing to own the entire ad stack on day one. A consumer brand plus an enterprise/API wedge creates two paths to scale: consumer subscriptions and usage-based revenue from companies that want grounded answers without building their own RAG stack. If they own the query interface and the citation graph, they compound data, brand, and distribution in a way that is hard to unwind. That is a multi-billion-dollar outcome even at modest share of search and knowledge work.

Product & wedge. Perplexity is an AI answer engine: users ask questions and get synthesized, cited responses with follow-ups, rather than a list of links. The wedge is grounded search—real-time web retrieval plus generation, with visible sources—positioned against both hallucinating chatbots and slow, ad-heavy SERPs. Consumer product (free + Pro) drives habit and brand; Pro adds higher limits, better models, file upload, and related features. Enterprise and API extend the same retrieval+generation loop to internal knowledge and developer use. Differentiation is UX speed, citation trust, and “search-first” rather than “chat-first.” The product is simple to try and hard to unlearn once it becomes the first place you ask a factual question.

Market & competition. TAM is internet search, knowledge work, and enterprise RAG/search—overlapping Google Search, ChatGPT, and vertical knowledge tools. Real competitors: Google (Search, AI Overviews, Gemini); OpenAI (ChatGPT with search/browsing); Microsoft/Bing Copilot; Anthropic (Claude with tools); xAI/Grok; You.com and other AI search startups; plus traditional SEO/content sites and internal enterprise search (Glean, etc.). Google has distribution, data, and ads; OpenAI has brand and model lead in many users’ minds. Perplexity’s bet is that “answers with sources, optimized for query completion” is a distinct product category, not a feature Google or OpenAI will fully absorb. Winner-take-most dynamics in default search make this a high-upside, high-risk category.

Traction & business signal. Publicly known: rapid consumer adoption of the answer-engine product; paid Pro tier; expansion into enterprise and API; high-profile investors and partnerships; continued product shipping (models, agents, shopping/related surfaces). Exact MAU, retention, paid conversion, ARR, gross margin, and enterprise ACV are unknown from public sources at the level a Series A memo would require. Fundraising and valuation history are public in later rounds but are not a substitute for unit economics or cohort data. Brand and usage anecdotes are strong; financial proof of a durable business is not fully public.

Risks (the three that kill the deal).

  1. Google (and OpenAI) copy the UX and keep the default. If AI Overviews + Gemini or ChatGPT search become “good enough” and sit on the browser/OS/default search slot, Perplexity remains a nice-to-have, not the habit. Distribution, not model quality, decides search.
  2. Unit economics of retrieval + generation. Every query is expensive (search + inference). If Pro conversion and enterprise pricing do not cover COGS at scale, or if ad/monetization lags, the company burns through capital while training users on a subsidized product. Unknown margins make this existential.
  3. Moat is thin: citations and UX are copyable. Without proprietary index depth, unique data rights, or a network effect beyond brand, they become a thin wrapper. Switching costs for consumers are low; enterprises will multi-home. Honest view: brand and speed are real but not yet a fortress.

Verdict: Invest at hypothetical Series A, conditional on one answer. The category is large, the product is the right wedge (grounded answers), and a consumer+enterprise push can create optionality. I would invest if the team can show they are not just a feature. The one question I need answered first: What is 90-day retention and paid conversion for cohorts that hit “aha” (e.g., 10+ queries in week one), and what is contribution margin per Pro/enterprise user after inference and retrieval cost? If retention and unit economics are real, this can be a fund-returner. If not, Google and OpenAI will compress them into a nice app with no durable economics.

DeepSeek V4 Pro vs Grok 4.6, overall

Across 29 judged battles, DeepSeek V4 Pro won 22, Grok 4.6 won 5, and 2 were ties. On investment memo alone: 7 to 2, 1 tied.

Compare DeepSeek V4 Pro and Grok 4.6 Judge one yourself, blind

Costs are the provider's list prices at the time of the run.