How Aravind Srinivas Built Google's Most Credible Challenger

There's a specific kind of ambition that grows up next to something great and refuses to be impressed by it.

Aravind Srinivas grew up in Chennai. His mother used to point at the IIT Madras campus on their bus rides, the same campus that sits right there in their neighborhood and tell young Aravind: "This is where you are going to study."

He got in. Graduated as the dual-degree topper in electrical engineering. Then left for Berkeley for his PhD. Then interned at Google Brain, DeepMind, and OpenAI, the three most consequential AI research labs in the world, before most people knew what a large language model was.

He saw exactly how the biggest companies on earth thought about search and information. He understood their architecture, their priorities, their blind spots.

And then, in August 2022, he left OpenAI and decided to build something none of them had bothered to build properly.

That decision made by a 28-year-old from Chennai who had spent his career inside the machine, became Perplexity AI. Today valued at $21 billion. Processing 300 million queries a week. With roughly 100 employees.

This is how it happened.

📌 No sponsor this week - Projects of AI just relaunched. If your product belongs in front of 2,400+ AI founders and builders with a 69% open rate, grab a slot here →

The boy who missed his branch change by 0.01

Aravind Srinivas was born on June 7, 1994, in Chennai, Tamil Nadu. Middle-class family. Father a retired banker. Mother a homemaker who had one fixed point on every bus ride: the IIT Madras campus near their home.

"This is where you are going to study."

He delivered. Got into IIT Madras, one of India's most competitive engineering institutions. But not in Computer Science. He got Electrical Engineering, and the branch change he needed required a CGPA above a certain threshold after the first semester.

He missed it by 0.01.

One-hundredth of a grade point. Depression followed for a year. And then in the way these stories always go, he found the thing he was actually supposed to find.

In his third year, stuck in Electrical Engineering, Srinivas discovered AI, machine learning, and neural networks. Not as a consolation prize. As a genuine obsession. He won the National Talent Search Scholarship. Got a merit certificate in the Indian Mathematical Olympiad. Graduated as the dual-degree topper.

The branch he didn't get into led him to the field that built the company. The 0.01 GPA miss that felt like a disaster became the detour that made everything.

Three labs. Three lessons. One blind spot.

After IIT Madras, Srinivas moved to UC Berkeley for his PhD in Computer Science, working under renowned AI researcher Pieter Abbeel. His doctoral thesis "Representation Learning for Perception and Control" covered deep learning, reinforcement learning, and robotics.

But the real education happened in the internships.

Google Brain: The scale of what was possible. How you build systems that serve billions of people. The infrastructure thinking that underpins everything Google does.

DeepMind, London: This is where the seed was planted. Srinivas's rental accommodation in London was so bad that he spent most nights sleeping at the office. And the office had a library.

One evening, exploring DeepMind's library, he found "In the Plex" a chronicle of Google's first 15 years by journalist Steven Levy. Srinivas devoured it, reading it multiple times. "That changed my whole understanding of startups," he says.

Think about the image for a second. A young researcher from Chennai, sleeping at his workplace in London because his flat is too grim to go home to, reading a book about Google's founding by the light of an AI research library. The seed of what would become a $21 billion company was planted in that room.

OpenAI: The humility check. "I went to OpenAI and I felt really bad because people were so much better than me," Srinivas recalled. "It was a big reality check that I could improve a lot more in programming, in first principles thinking, in my clarity of thought."

He improved. Got hired as a full-time research scientist at OpenAI after his PhD. And then, in August 2022, he left.

The founding insight that sounds obvious in hindsight

The idea behind Perplexity sounds simple now. It wasn't obvious in 2022.

The fundamental insight: Google was broken not technically, but experientially. You type a query. Google gives you ten links. You spend 20 minutes clicking through pages, trying to find an answer. The search engine doesn't answer, it redirects.

Srinivas and his co-founders wanted to build something different. An answer engine, you ask a question, you get a direct, concise, cited answer synthesized from real-time web sources. Not ten blue links. An answer.

The team he assembled was exceptional in its complementarity:

• Denis Yarats (CTO): Former AI researcher at Meta's Fundamental AI Research team. Deep technical research background, Srinivas's equivalent on the systems side.

• Johnny Ho (CSO): Former engineer at Quora with deep experience in search and ranking. The person who understood search product from the inside.

• Andy Konwinski: Co-founder of Databricks, one of the most successful data infrastructure companies ever built. The person who'd scaled something from zero before.

They founded Perplexity AI, Inc. in August 2022. Then, seven days after ChatGPT launched, on December 7, 2022, Perplexity went live.

Seven days. ChatGPT changed the entire public conversation about AI and Perplexity was already live.

The decision that defined everything

Here's the strategic call that most people overlook when they tell the Perplexity story.

They decided not to build their own large language model.

In 2022 and 2023, every serious AI startup was racing to build foundation models. Billions were being poured into training runs. The implicit assumption across the industry was that you needed your own LLM to be a real AI company.

Srinivas and his team looked at that race and made a different bet. They didn't need to build the model, they needed to build the experience on top of the model. The search interface, the citation system, the answer synthesis, the real-time web access. That's where the value was. That's what Google hadn't built. That's what they could win.

"It was a decision driven through conviction and pragmatism," Srinivas explained later.

The result: while other startups burned through hundreds of millions training LLMs, Perplexity used existing models as a layer and focused every engineering hour on what it actually felt like to get an answer. That focus is why the product felt different from day one.

Not building your own LLM was the most important product decision Perplexity ever made.

What happened after launch - in numbers

The growth from December 2022 to today is one of the most compressed valuation trajectories in the history of technology.

December 2022: Perplexity launches. First public users.
March 2023: $26M Series A. Valuation: $150M. Led by NEA, with Jeff Bezos participating personally. January 2024: Valuation hits $520M.
April 2024: $1 billion valuation. Unicorn in 16 months.
May 2025: $500M Series E. Valuation: $14–18B. Nvidia, SoftBank, IVP, Accel all in.
September 2025: $200M raise. Valuation: $20B.
January 2026: Series E-6. Valuation: $21.21B. Total raised: $1.72B across 11 rounds.

Revenue followed the same trajectory: from near zero in early 2023, to $148M ARR by mid-2025, to $500M annualized revenue by April 2026, up 335% year-over-year, driven by the February 2026 launch of its AI agent product and a new usage-based pricing layer.

The company serves more than 100 million monthly active users with roughly 250 people, generating approximately $2 million ARR per employee.

Perplexity now processes more than 300 million queries per week

In October 2025, Aravind Srinivas debuted on the M3M Hurun India Rich List with an estimated net worth of $2.5 billion. At 31, he became India's youngest billionaire.

The kid who missed his CS branch change by 0.01 GPA points.

The Google connection nobody talks about enough

There's a detail in this story that keeps pulling at me.

Srinivas grew up in Chennai idolising Sundar Pichai, the current CEO of Google because Pichai also grew up in Chennai. Same city. Similar middle-class background. Engineer who left India, went to an American university, and built something enormous.

The company Srinivas is building is the most serious challenger Google's search business has faced in twenty years.

He's not just competing with Google. He's following a template set by someone from his hometown and trying to do what that person never did, which is challenge the thing that person built.

In February 2026, Perplexity killed its advertising strategy entirely, going subscription-first to protect user trust. A bold call given that search advertising is one of the most profitable business models ever created.

Pichai's Google is built on ads. Srinivas's Perplexity is built on subscriptions. Two people from Chennai. Two completely different bets on what the internet's information layer should look like.

Where Perplexity is now

The product has expanded well beyond search. Perplexity now offers:

• Perplexity Pro ($20/month): Access to GPT-4, Claude, Gemini, and Perplexity's own models plus PDF analysis, unlimited queries, and internal knowledge search

• Perplexity Max ($200/month): AI email assistant, unlimited Labs access, priority support

• Perplexity Computer: An AI agent that can browse the web and take actions launched February 2026, immediately driving the ARR spike

• Perplexity Comet: An AI-powered browser (in development)

The company committed $750 million to Microsoft Azure infrastructure in January 2026, an aggressive signal of how seriously they're scaling.

They've also struck distribution partnerships with Bharti Airtel in India, giving them millions of users overnight and they bid for both TikTok US and Google Chrome when regulatory pressure made those assets available. Both bids were unsuccessful, but both signalled the same thing: Srinivas isn't thinking about building a better search box. He's thinking about owning the layer through which people access information entirely.

Management's 2026 revenue target is $656 million implying continued compounded monthly growth from today.

The answer engine that launched seven days after ChatGPT is now targeting $656 million in annual revenue three and a half years later.

What I keep coming back to

The Perplexity story has a lot of elements that make it easy to tell as a "brilliant founder executes perfectly" narrative. And Srinivas is clearly exceptional.

But the detail I keep coming back to isn't the strategy or the growth numbers. It's the 0.01 GPA.

The branch change he missed by a fraction of a point felt, at 18, like a disaster. He spent a year depressed about it. And then, stuck in Electrical Engineering, he found AI, the field that he turned into a $21 billion company.

The thing that didn't go according to plan became the plan.

That's a pattern worth sitting with, whatever you're building right now.

That's the Perplexity origin story. On Friday I'm going inside the product — the answer engine vs. the search engine, why Google still hasn't properly responded, and what builders can steal from how Perplexity won without building their own LLM. See you then.

Hit reply and tell me - did you know about the 0.01 GPA story? I read every reply.

- Projects of AI

P.S. - Srinivas's childhood hero was Sundar Pichai, the Google CEO who also grew up in Chennai. The company Srinivas built is now Google's most serious search challenger in two decades. I don't know if that's poetic justice or just Chennai producing exceptional people. Probably both.

Enjoying this? Forward it to one person building in AI. That's the only growth strategy I have.

Want to reach 2400+ AI founders and builders? Sponsor Projects of AI