Companies that can provide the greatest economic value through the power their AI uses will ultimately achieve the highest valuations, Perplexity CEO Aravind Srinivas told CNBC on Wednesday.
Srinivas said whichever company can deliver “the highest value per watt per user” will be the winner in the future.
“Whoever is able to maximize that particular goal will really do it, balancing accuracy, latency, cost, privacy and intelligence, they’re going to win, that’s what’s going to win in the long run,” Srinivas told CNBC. Elaine Yu in an interview Wednesday.
A token refers to the basic unit of data that an AI model can process. When an AI chatbot is asked to perform a task, it breaks it down into tokens. Each token then requires energy to process. According to Srinivas, whichever company can provide the best ratio of energy to economic output will be in the strongest position.
“So it might seem like some model providers are making a lot of money because their models are very expensive… but that’s short-term revenue growth,” Srinivas said.
Perplexity is more interested in agentic AI, a term that refers to AI systems that can handle more complex tasks beyond simple queries. In February, the company announced Perplexity Computer, an agent capable of running complex tasks over long periods of time.
Although Perplexity develops some of its own models, its key products incorporate models from other AI companies like Anthropic. One of Perplexity’s main goals is to improve efficiency to achieve the best results while minimizing energy consumption. To support this goal, Perplexity on Tuesday announced Personal Computer, a tool it calls an “orchestrator.”
The orchestration process involves a system that makes decisions about the best model to use for a particular task, how agents work together, and where the AI should process requests. Today, much of the actual processing of AI is done in data centers.
But AI companies are increasingly working to enable these models to be processed on a device like a phone or laptop. Experts say this could reduce the power needed for AI processing, making it faster and more secure, since the data is not sent to a server. Perplexity Personal Computer automatically routes processing where it sees fit.
“The data center is coming to your laptop,” Srinivas said, adding that it is crucial to have an AI operating system that brings everything together into one unified system.
On Wednesday, Perplexity announced that its Personal Computer product would be available on Microsoft’The Windows operating system allows AI to connect to applications such as Word and Outlook, as well as files on a user’s device. Perplexity has already launched the personal computer on AppleThe Mac product from.
Srinivas said Perplexity is focused on creating a “sustainable, sustainable advantage” over its competitors and that “this is an orchestration problem.”
“We believe that by solving this problem, we will build a high-value business with long-term sustainable advantage,” Srinivas said.
Growing competition Perplexity faces increasing competition as rivals like OpenAI, Anthropic, and Google focus more on AI agents.
Perplexity, last reported to be valued at $20 billion, lags behind Anthropic and OpenAI, whose valuations have soared to almost 1,000 billion dollars and just more than 850 billion dollarsrespectively. Anthropic this week filed confidentially for a U.S. IPO as investor demand for AI shares continues.
Even as Perplexity expands to Microsoft and Apple products, those same companies are developing their own AI agents. On Tuesday, Microsoft announced a new coding and reasoning models. Apple currently working on a updated version of its digital assistant Siri based on Google AI models.
Srinivas said that while these companies will build their own AI, Perplexity’s platform-agnostic approach will help it compete.
Learn more
“I think they’re absolutely going to try to build their own AI systems, but we think we’re building the most versatile operating system by running it on different models, on different chips, on different legacy operating systems, different hardware vendors, different laptops,” Srinivas said.
“This neutral, hybrid orchestration layer is what we do, and it allows us to balance all the different objectives simultaneously.”
Every time Anthropic’s models improve, Perplexity also improves because Anthropic’s models are integrated into Perplexity, the CEO said. This has led Perplexity to triple its annualized revenue since the start of the year, “thanks to the model advancements made by Anthropic,” he told CNBC.
