Perplexity CEO Aravind Srinivas told CNBC on Wednesday that companies that can deliver the most economic value from AI-enabled capabilities will ultimately command the highest valuations.
“The companies that can deliver the most value per watt per user will be the winners in the future,” Srinivas said.
“Whoever can best accomplish this specific objective by balancing accuracy, latency, cost, privacy and intelligence will win. That’s the long-term winner,” Srinivas told CNBC’s 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 into tokens. Processing each token requires energy. Srinivas’ view is that companies that can provide the optimal ratio of energy to economic output are in the strongest position.
“Some model providers may feel like they’re making a ton of money because their models are so expensive…but it’s a short-term revenue boost,” Srinivas said.
Perplexity is increasing its focus on agent AI, a term that refers to AI systems that can go beyond simple queries and handle more complex tasks. In February, the company announced Perplexity Computer, an agent that can perform complex tasks over long periods of time.
Although Perplexity develops some of its own models, its main product integrates models from other AI companies such as Anthropic. Perplexity’s main focus is improving efficiency to achieve the best results while minimizing energy usage. To help with that goal, Perplexity on Tuesday announced Personal Computer, a tool it calls an “orchestrator.”
The orchestration process includes systems that determine the best model to use for a particular task, how agents work together, and where AI processes queries. Currently, much of the actual AI processing takes place in data centers.
But AI companies are increasingly focused on making these models processable on devices like phones and laptops. Experts say that because data is not sent to servers, less power is needed to process the AI, which could make it faster and more secure. Perplexity Personal Computer automatically routes processing where it thinks it’s best.
“The data center is coming to the laptop,” Srinivas said, adding that an AI operating system that brings everything together into a single integrated system is key.
Perplexity announced Wednesday that its Personal Computer product Microsoft Corporation■ Windows operating system. Enable AI to connect to apps like Word and Outlook and files on your device. Perplexity already sells personal computers. appleMac products.
Srinivas said Perplexity is focused on creating a “sustainable and lasting advantage” over competitors, and that “this is a matter of orchestration.”
“By solving that, we believe we can build a very valuable company with lasting, long-term advantages,” Srinivas said.
Intensifying competition
Perplexity faces increased competition as rivals such as OpenAI, Anthropic and Google increase their focus on AI agents.
Perplexity, which was last reported to be valued at $20 billion, trails behind Anthropic and OpenAI, whose valuations have risen to nearly $1 trillion and just over $850 billion, respectively. Anthropic secretly filed for an initial public offering in the U.S. this week as investor demand for AI stocks continues.
While Perplexity is expanding its offerings across products from Microsoft and Apple, those very same companies are developing their own AI agents. On Tuesday, Microsoft announced a new coding and inference model. apple is currently working on an updated version of its digital assistant Siri, based on Google’s AI models.
Srinivas said these companies build their own AI, but Perplexity’s platform-agnostic approach will help it compete.
“I think they’re definitely trying to build their own AI systems, but we believe we’re building the most versatile operating system by making it work on different models, different chips, different legacy operating systems, different hardware providers, different laptops,” Srinivas said.
“What we are working on is a hybrid, neutral orchestration layer that allows us to balance all of our different objectives at the same time.”
Anthropic’s models are integrated into Perplexity, so whenever Anthropic’s models improve, Perplexity also improves, the CEO said. This allows Perplexity to triple its annual revenue since the beginning of the year “thanks to the model advancements made by Anthropic,” he told CNBC.
