Aravind Srinivas

Who they are

Aravind Srinivas is co-founder and CEO of Perplexity AI — a Berkeley PhD who interned at DeepMind, OpenAI, and Google before founding Perplexity in 2022 and angel-investing in Cursor, Eleven Labs, and Mistral.

Person

Aravind did a dual degree in Electrical Engineering at IIT Madras, then crossed to Berkeley for a PhD in Computer Science, finishing in 2021. That academic track took him through research internships at Google, DeepMind, and OpenAI — the three labs that defined the first wave of deep learning — before he landed a Research Scientist role at OpenAI full-time. He co-founded Perplexity AI in August 2022, building it from scratch into an answer engine processing over 1 billion queries monthly by mid-2026. The through-line is a researcher who decided the most interesting experiment was a company: every move from IIT to Berkeley to the lab circuit pointed toward building something at the frontier, not just studying it. He angel-invests actively — Cursor, Eleven Labs, Suno, Mistral, Cognition, Lovable, Fal, LiveKit, and Pika — which means he is wired into the early-stage AI ecosystem well beyond his own company. He speaks publicly on AI product strategy, the value of fear as a competitive motivator, and the infinite value of knowledge, with appearances at Stanford GSB, Harvard SEAS, and the Joe Rogan and Harry Stebbings podcasts in 2026.

Company

The most recent move: in February 2026, Perplexity dropped AI-integrated advertising entirely and shifted to a subscription-first model, citing user trust and result objectivity. That followed a $200 million funding round in June 2026, which brought total funding to approximately $1.72 billion and a valuation of $22.6 billion as of January 2026. In January 2026, Perplexity signed a three-year $750 million commitment to Microsoft Azure to support infrastructure and federal procurement — and in November 2025 inked a direct-to-government deal with the U.S. General Services Administration to accelerate AI adoption in federal agencies. The product surface is expanding fast: in 2026, Perplexity launched the Comet AI-powered browser globally on iOS, Android, Mac, and Windows, and earlier launched Model Council — running GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro in parallel so users can see where models agree and disagree. Srinivas has ruled out an IPO before 2028, targeting $656 million ARR for 2026 against approximately $450 million ARR reached by mid-2026.

Market

Perplexity sits in the AI search and conversational answer market, directly challenging Google's dominance alongside Microsoft's Bing-OpenAI integration, OpenAI's ChatGPT, Anthropic's Claude, Genspark, and Grok. It holds between 6% and 8% of the AI chatbot market share in 2026, with over 45 million monthly active users, and positions itself on transparency and trust at a moment when Google is deepening Gemini integration into Chrome. The regulatory environment is turbulent: Perplexity faces a lawsuit from CNN over unauthorized use of content, broader copyright infringement actions from publishers, and antitrust scrutiny of Google — the same scrutiny that prompted Perplexity to propose a $34.5 billion acquisition of Google's Chrome browser.

Network

Aravind's inner circle at Perplexity includes three technical co-founders — Andy Konwinski, Denis Yarats, and Johnny Ho — who lead the engineering and research side of the company. On the business side, Dmitry Shevelenko serves as Chief Business Officer; he has since joined the board of Lazard, a signal of Perplexity's growing institutional profile. Aravind is also closely tied to Harry Stebbings (20VC) in the media and investor network, having done what Stebbings described as one of his standout interviews in 2026.

Andy Konwinski· Co-founder, Perplexity AI
Denis Yarats· Co-founder, Perplexity AI
Johnny Ho· Co-founder, Perplexity AI
Dmitry Shevelenko· Chief Business Officer, Perplexity AI; Board Member, Lazard
Harry Stebbings· Host, 20VC podcast

How they likely show up

Research internships at Google, DeepMind, and OpenAI before founding — suggests someone who stress-tests ideas across institutions before committing, not an instinct-first operator.
Founded Perplexity immediately after his PhD rather than staying in research — high agency, low tolerance for not owning the outcome.
Angel-investing in nine early-stage AI companies (Cursor, Eleven Labs, Mistral, etc.) alongside running a $22.6 billion company — runs multiple parallel bets, comfortable with context-switching at speed.
Public content themes center on fear as motivation and AI product strategy — likely responds well to direct competitive framing rather than consensus-building language.
Dropped advertising revenue in February 2026 in favor of subscriptions — willing to sacrifice near-term revenue for a long-term strategic position, not a quarter-to-quarter thinker.
Ruled out IPO before 2028 despite a $22.6 billion valuation — prioritizes product depth over liquidity events; probably impatient with conversations that optimize for exit rather than product.

Conversation tips

→ Reference the Model Council launch specifically — running three frontier models in parallel to show disagreement is an unusual product bet that reveals his thinking on AI transparency.
→ Ask about the advertising-to-subscription shift in February 2026 — it was a deliberate revenue trade-off and he'll have a strong view on why it was the right call.
→ He speaks publicly about embracing fear of competitors stealing ideas as a motivator — lean into competitive tension, don't soften it.
→ His angel portfolio (Cursor, Mistral, Eleven Labs) is a window into where he thinks AI is going; asking about a specific portfolio company will signal you've done the work.
→ Don't generalize about 'AI search' as a category — he's differentiated Perplexity sharply from chatbots and from Google; he'll want precision.

Toolbox

Openers

Open on Perplexity Brain — he announced it as a self-improving context-graph system for persistent agentic memory, which is a specific architectural bet about where AI assistants go next.
Reference SPACE, Perplexity's internally built sandbox lifecycle management platform for agent compute environments — he posted about it publicly, and it signals the company is building infrastructure most search companies would outsource.
Lead with the Model Council launch — running GPT-5.4, Claude Opus 4.6, and Gemini 3.1 Pro in parallel to show users where models disagree is a sharp product philosophy statement about AI transparency.

Discovery questions

You dropped advertising in February 2026 to go subscription-first — how does that reshape what you build for, and what did you have to give up to make that call?
The government deal with GSA and the $750 million Azure commitment happened close together — is the enterprise and federal track a deliberate second surface, or did it emerge from inbound?
You angel-invested in Cursor, Mistral, and Cognition among others — where do you see the frontier moving that Perplexity isn't positioned to capture directly?

Avoid

Don't open with broad AI industry trends or regulatory commentary — he engages with specific product and competitive bets, not category-level framing.

Make it yours