Artificial Intelligence · AI Search

Perplexity

Builds an AI-native answer engine combining web retrieval and generative models.

Research status: Research coverage
Headquarters: San Francisco, CA
Coverage: 4 product lines mapped
Last reviewed: September 2026

Product map

What the company builds

Products and operating platforms help connect company-level news to the revenue pools, customers, competitors, and supply chains it may affect.

Answer Engine

Citation-led conversational search and research.

Perplexity Pro

Premium research and model access.

Enterprise Pro

Knowledge search and research for organizations.

Comet

AI-native web browser and agent experience.

01 / THE BUSINESS

How Perplexity creates value.

Builds an AI-native answer engine combining web retrieval and generative models. The research focus is whether customer adoption can support attractive economics after development, delivery and ongoing service costs.

SectorAI

Founder coverage3 profiles

Research checkedSep 18, 2026

02 / PEOPLE BEHIND THE COMPANY

Meet the builders.

Co-founder

Aravind Srinivas

Aravind Srinivas is a co-founder of Perplexity. An AI researcher who helped build Perplexity around search, retrieval and generated answers.

Explore founder

Co-founder

Denis Yarats

Denis Yarats is a co-founder of Perplexity. The company’s founding and leadership material provides context for its work in ai search.

Explore founder

Co-founder

Johnny Ho

Johnny Ho is a co-founder of Perplexity. The company’s founding and leadership material provides context for its work in ai search.

Explore founder

Selected founders; historical founding roles are distinct from current management positions.

03 / CAPITAL & OWNERSHIP

Follow the financing.

Selected disclosed events, newest first. Announcement values describe that transaction and date; they are not current share prices. Debt, equity and secondary transactions are identified separately.

A financing figure is not established here.

Use the original company sources below to check new announcements. We have not substituted an estimate for a disclosed round.

04 / IN THEIR OWN WORDS

Watch. Listen. Form your view.

Lex Fridman Podcast · 2024-06-19

Aravind Srinivas — AI, search and the internet

The Perplexity co-founder discusses search and the development of the answer engine.

Watch on YouTube ↗

Open with publisher ↗

Perplexity

Perplexity — company videos

Product demonstrations, company presentations and original updates. Browse the channel for the latest available material.

Visit original material ↗

Interviews reflect the speaker’s perspective at publication. Company statements are not independent verification.

05 / EDITORIAL ANALYSIS

The opportunity. The open questions.

Our interpretation of the business model and cited sources, not a valuation or a recommendation.

What could work

The case to investigate

Consumer search behavior, distribution, and the economics of AI-generated answers.
A stronger product position becomes economically meaningful when customers renew, expand and pay enough to cover delivery costs.

What could go wrong

Pressure-test the thesis

Incumbent competition.
Content licensing.
Inference costs.

What to watch next

Query and subscriber growth — what changed in the latest original disclosure?

Distribution agreements — what changed in the latest original disclosure?

Advertising and commerce products — what changed in the latest original disclosure?

YOUR RESEARCH WORKSPACE

Keep the questions that matter.

Use this checklist to record what you have investigated. Your notes stay in this browser.

Query and subscriber growth — what changed in the latest original disclosure?Distribution agreements — what changed in the latest original disclosure?Advertising and commerce products — what changed in the latest original disclosure?

My notes on Perplexity

06 / RESEARCH TRAIL

Go straight to the source.

Company overviewperplexity.ai Founding & leadership backgroundperplexity.ai

Research checked Sep 18, 2026. Selected public disclosures, not a complete capitalization table or securities-filing search. Missing information is left unconfirmed. No investment availability or IPO date is implied.

Why SMYC is watching

Consumer search behavior, distribution, and the economics of AI-generated answers.

Private-company information is incomplete, episodic, and difficult to verify. Treat every data point as a starting point for diligence—not a substitute for company-approved materials, legal review, or investment analysis.

Add to my private current

Signals to monitor

Query and subscriber growthDistribution agreementsAdvertising and commerce products

Risks to underwrite

Incumbent competitionContent licensingInference costs

Underwriting map

What an investor should actually diligence

Separate recurring software revenue from usage-based inference, services, consumer subscriptions, licensing, and strategic financing. Model revenue quality after compute and distribution costs.

Buyer map

Developers and technical teams

Enterprise business units

Regulated and security-sensitive organizations

Consumers and prosumers

Economics to request

Revenue and annualized run-rate by product

Gross margin after inference and hosting

Net revenue retention and cohort expansion

Compute commitments, utilization, and vendor concentration

Sales efficiency and support burden

Moat tests

Model quality persists after price normalization

Proprietary data or workflow context improves retention

Distribution is owned rather than rented

Switching costs survive model commoditization

Milestone map

Major model or product release

Enterprise deployment expansion

Material distribution partnership

Improving inference efficiency

Comparable lensesFoundation-model platformsCloud AI servicesVertical AI applicationsDeveloper infrastructure

Underwriting framework

How to diligence Perplexity

A company-specific starting point based on the operating realities of artificial intelligence. Replace assumptions with verified company materials, customer evidence, legal documents, and expert work.

Economics to model

Annualized recurring revenue versus contracted value

Inference gross margin after compute

Net revenue retention by customer cohort

Research and compute spend per dollar of growth

Evidence of proof

Production workloads rather than pilots

Usage depth across teams and workflows

Model performance on customer-specific tasks

Renewals without unsustainable discounting

Comparable lenses

Foundation-model providers

Cloud AI platforms

Application-layer AI companies

Open-source and in-house substitution cost

Questions to answer

Is differentiation in models, data, distribution, or workflow?

How quickly does capability commoditize?

Who controls the customer and the compute bill?

What safety or regulatory event could slow adoption?

No securities are offered here.

SMYC does not represent that shares are available, transferable, appropriately priced, or suitable for any person. Private securities are illiquid, speculative, subject to transfer restrictions, and may result in total loss.