Perplexity is not following the AI IPO trend: it will only consider an IPO in 2028, first developing the company into a "healthy, high-growth" business.

As OpenAI and Anthropic race to go public one after the other, Perplexity CEO Aravind Srinivas chooses to take a contrarian approach, clearly stating that the company plans to go public only around 2028, and that this timetable will not change due to competitors' IPO actions.
(Background: OpenAI has already filed for an IPO! Following Anthropic and SpaceX, they are making their move on Wall Street.)
(Additional context: Perplexity has launched a hybrid inference engine, keeping sensitive data local, while complex reasoning runs in the cloud.)

Table of Contents

Toggle

• Giants racing to go public, Perplexity deliberately absent
• Using spare change to generate $450 million ARR
• Is small and patient an advantage or the only option?

In this wave of AI IPOs, SpaceX, Anthropic, and OpenAI have all submitted IPO applications. However, at this moment, Perplexity CEO Aravind Srinivas told CNBC, “It has nothing to do with these two companies not going public; Perplexity has always planned to IPO around 2028, and that hasn’t changed.”

Giants racing to go public, Perplexity deliberately absent

OpenAI and Anthropic nearly simultaneously filed confidential IPO documents with the U.S. SEC, with both companies valued close to the trillion-dollar mark.

In contrast, Perplexity’s figures seem stark: its most recent funding round in September 2025 raised $200 million, with a post-money valuation of about $20 billion, and total funding reaching approximately $900 million. OpenAI’s single round raised $122 billion, and Anthropic’s single round raised $30 billion. Just one round for OpenAI exceeds Perplexity’s valuation by more than six times.

Faced with this reality, Srinivas chooses to position himself as a “spectator.” He defines the next six months as a “market test period” for AI giants going public, where the public markets will reveal whether these billion-dollar valuations can hold up. His stance is: if the public reaction is poor, it’s a problem for OpenAI and Anthropic, not for Perplexity, because Perplexity is not even in this wave.

Is this confidence, or just a way of saying it?

Using spare change to generate $450 million ARR

Srinivas’s confidence in waiting stems from a set of numbers: Perplexity’s ARR (Annual Recurring Revenue) surpassed $450 million in March 2026, and the company aims to reach $656 million by the end of the year.

Driving this acceleration is a new product: Perplexity Computer. This is an autonomous agent platform that integrates multiple AI models into a unified research workflow, allowing users to avoid switching between different AI tools. Perplexity coordinates the models behind the scenes to complete tasks. Within 30 days of launch, ARR growth accelerated by about 50%.

From a capital efficiency perspective, this figure is rare. With approximately $900 million in total funding, achieving an annualized revenue scale of $450 million means that every dollar raised generates about $0.50 in ARR. This ratio is highly efficient among current AI startups. OpenAI and Anthropic’s investments in compute power, model R&D costs, and talent wars all require ongoing large-scale funding, and they must go public partly to access more public market capital.

Perplexity’s strategic logic is: deliberately setting 2028 as the earliest IPO date, with the goal of cultivating a “healthy, high-growth” business rather than being forced to go public to satisfy investors’ exit timelines. Srinivas’s tone carries a clear rhythm: we know where we’re headed, and we don’t need to be disrupted by someone else’s schedule.

Is being small and patient an advantage or the only option?

But here’s a question I can’t answer for Srinivas.

Perplexity indeed demonstrates a different path: building a substantial revenue-generating AI application layer with far less capital than competitors. Early data from Perplexity Computer supports this approach—integrating AI agents into workflows offers higher commercial conversion value than simple search and Q&A.

However, the line between “choosing to wait until 2028” and “only able to wait until 2028” is subtle. If, after going public, OpenAI and Anthropic’s valuation multiples are significantly compressed, it could impact the later-stage financing environment for the entire AI startup ecosystem. When that time comes, Perplexity’s capital market environment in 2028 may not be more friendly than it is now.

Another variable is the competitive landscape. Both OpenAI and Anthropic are accelerating their deployment of AI agents and workflow products, which is currently Perplexity Computer’s main battlefield. Capital advantages in application layers are equally effective—this is the other side of “compute power determines the pattern, capital determines the speed.”

Srinivas says he believes patience is an advantage, but this judgment remains an unproven hypothesis. Over the next six months, this test will not only be for AI giants but also for Perplexity.