Why is Perplexity AI so famous?

The Search Frustration Perplexity Solved

You type a question into Google, get ten blue links, open five tabs, and fifteen minutes later you still don't have an answer. That friction — repeated billions of times a day — is exactly the gap Perplexity AI was built to close.

Traditional search engines return a wall of links and leave the synthesis work to you. Chatbots, meanwhile, answer fluently but often hallucinate without sources, forcing you to verify every claim yourself. Perplexity positioned itself as neither: an "answer engine" that reads live web sources and delivers conversational answers with inline, clickable citations, as industry analysis and its own product documentation describe.

The numbers suggest the concept works. An independent 100-search head-to-head test found Perplexity beat Google 61–29, with 10 ties — dominating research and synthesis questions 31–2. The tester estimated 2–3 hours saved across 35 research queries alone. As she put it: "Perplexity reads multiple sources and synthesizes an answer with citations. Google gives you a list of links and hopes you'll figure it out."

What the answer-engine model changes for a typical research task:

• Synthesis over links — multiple sources distilled into one cited answer instead of ten tabs
• Verifiability — inline citations let you check any claim in one click
• Follow-up conversation — refine the question without restarting the search
• Zero friction — free web access with no registration required

That last point matters more than it sounds. By staying free and open without sign-up, Perplexity removed every adoption barrier, helping monthly users climb from 2 million to roughly 30 million in about two years, according to Search Engine Land's statistics guide.

One honest caveat: citations speed up verification, they don't replace it. As the same tester noted, "citations don't mean the synthesis is correct" — they make checking faster, not unnecessary. That trust-but-verify instinct is exactly why independent benchmarking matters, whether for search tools or the agent skills plugged into your workflow. It's the same principle SkillzBench is built on: free, verified, security-scanned skill benchmarks, so you know what actually works before you rely on it.

The Growth Story: From 2 Million to 45 Million Users

In April 2023, Perplexity AI was a scrappy startup valued at $121 million with 2 million monthly users. Two years later, it commanded a valuation above $20 billion — a 165x increase in roughly 30 months — and served tens of millions of people every month, according to industry tracking.

The user arc tells the same story. Perplexity hit 10 million monthly active users in January 2024, roughly 30 million by April 2025, and 45 million in late 2025, with search-industry data also showing 800% year-over-year user growth. Query volume climbed in parallel, from 400 million monthly searches in October 2024 to around 780 million by May 2025.

Several factors made this adoption curve possible:

• Zero-friction access. The web version is free and requires no registration, removing the single biggest adoption barrier most AI tools face.
• Retention. The platform reports an 85% retention rate, suggesting users who try it tend to stay.
• Distribution deals. An Airtel partnership in India drove a 640% user increase in one quarter, plus 2.8 million app downloads.
• Mobile momentum. Perplexity passed 80.5 million lifetime app downloads, peaking at 15.5 million monthly installs in October 2025.

That last point hints at a subtle shift in how people use the product. Website visits actually declined 13.21% month-over-month in November 2025, even as app installs peaked — evidence that users are moving from casual web queries to the app as a daily research habit, per usage statistics.

The money followed the users. Annualized revenue grew from under $250 million to more than $750 million within a single year, and Reuters reported Nvidia in talks to invest at a $30 billion-plus valuation. With roughly 100 employees, that works out to about $2 million in annualized revenue per employee — remarkable efficiency for a company in the research and analysis category.

Growth at this speed always invites scrutiny, and Perplexity's fame is partly built on controversy too — publisher lawsuits and a privacy class action (allegations not proven in court) generated as many headlines as the product itself. But the underlying bet is about trust: the company dropped advertising entirely in early 2026, with an executive telling the Financial Times that "the challenge with ads is that a user would just start doubting everything."

It's the same logic that separates useful AI tooling from risky tooling. Whether you're choosing an answer engine or the agent skills you plug into your workflow, verification matters — which is exactly why SkillzBench is building a free directory that security-scans, version-pins, and benchmarks every skill so you know which ones actually help. Join the waitlist to be first in line when it launches.

Headlines, Backers, and Controversy: The Fame Multipliers

A product can be genuinely good and still be ignored — Perplexity has made sure that never happens. Beyond the answer engine itself, a steady drumbeat of celebrity investors, audacious bids, and public feuds has kept the company in headlines almost continuously.

The backer list reads like a tech-world all-star roster. Investors include Nvidia, Jeff Bezos, SoftBank, Accel, and NEA, and Nvidia has reportedly been in talks to invest at a valuation exceeding $30 billion — a more than 50% jump year-over-year, per Reuters reporting. Cristiano Ronaldo added mainstream star power in December 2025, taking an undisclosed stake and brand partnership.

Then came the bids that generate headlines by design:

• A January 2025 proposal to merge with TikTok US
• A $34.5 billion offer to buy Chrome from Google in August 2025
• A contract to build a chatbot for Truth Social

Distribution deals turned attention into users. A Snapchat integration reaching roughly 1 billion users launched in January 2026, while an Airtel partnership in India drove a 640% user increase in Q2 2025 and 2.8 million app downloads in a single quarter. Gannett Media even announced a partnership for its USA TODAY Network, lending mainstream media legitimacy.

Controversy, though, cuts both ways. Legal disputes with the New York Times, Dow Jones, BBC, Forbes, Reddit, and Japanese newspapers have generated enormous coverage, and the Cloudflare clash got personal — Cloudflare's CEO tweeted that Perplexity acts "more like North Korean hackers," which Perplexity dismissed as a "charlatan publicity stunt." A March 2026 class action alleges the company shared users' chat data, including in "incognito mode," with Meta and Google via tracking pixels — allegations that remain unproven in court.

The lesson is clear: visibility compounds. Each lawsuit, bid, and celebrity stake produces coverage that feeds the next round of curiosity. The same trust question applies to the AI tools you bolt onto your own workflow — Perplexity's February 2026 pivot to subscriptions was explicitly about preserving user trust, and SkillzBench is building its free skills directory around the same principle: security-scan and benchmark every agent skill before you rely on it.

The Trust Bet — and What It Means for Your AI Stack

In February 2026, Perplexity quietly killed its advertising business. The reason, as one executive told the Financial Times, was blunt: "the challenge with ads is that a user would just start doubting everything." It was a deliberate bet that trust, not ad revenue, is what makes an answer engine worth paying for.

The numbers suggest the bet was affordable — and possibly necessary. Advertising had generated roughly $20,000 of Perplexity's $34M total 2024 revenue, a rounding error by any measure. Going subscription-first let the company prioritize "objective results over ad revenue" without sacrificing much income.

That trust positioning pairs with measurable quality. Perplexity reports 94% overall factual accuracy and 97% citation accuracy on benchmarks, which is central to why users treat its answers as reliable. Even independent testers who praise its head-to-head win over Google (61–29 across 100 real searches) add a caveat worth remembering: citations make verification faster, not unnecessary.

Now Perplexity is extending that trust equation into agentic AI. Its Comet browser assistant and cloud-based "Perplexity Computer" agent are credited with part of the company's revenue growth, and a Perplexity/Harvard study of hundreds of millions of anonymized interactions found that 57% of agent activity is cognitive work — users treating agents as thinking partners rather than task robots.

Here's why this matters for your own AI stack: every agent you deploy runs on instructions and skills, and those carry the same trust risks Perplexity engineered away. Published research found that ~26% of agent skills carry at least one vulnerability — and unlike a search result, a compromised skill acts on your behalf inside your workflow.

When you're choosing skills for Claude Code, Codex, Cursor, or Gemini CLI, apply the same logic Perplexity did. Ask:

• Can I verify exactly what version of this skill I'm running?
• Has it been security-scanned, or am I trusting an anonymous upload?
• Does it actually improve my agent's output, or just add instructions?
• Is the evaluation independent, or does the seller grade its own homework?

That last question is where most tools fail today. SkillzBench is building a free directory that pins every skill to a commit SHA, security-scans it, and benchmarks the Lift in a sandbox — so you can see, in percentage points, which skills genuinely help. Trust scores and Lift are always published separately, and money never touches either.

Perplexity wagered that users pay for answers they don't have to doubt. The same wager applies to the skills your agents run on: verify first, then trust.