Perplexity CEO tells CNBC one metric will determine who wins the AI race
Perplexity CEO Aravind Srinivas told CNBC on Wednesday that companies that can deliver the most economic value from the power of AI will ultimately achieve the highest values.
Srinivas said the company that provides “the most received value per watt per user” will be the winner in the future.
“Whoever manages to maximize that particular goal, balancing accuracy, latency, cost, privacy and intelligence all together, that’s what’s going to win in the long run,” Srinivas told CNBC’s Elaine Yu in an interview on Wednesday.
Token refers to the basic unit of data that an artificial intelligence model can process. When an AI chatbot is asked to perform a task, it divides it into tokens. Each token requires energy to process. Srinivas’ view is that the company that can provide the best ratio of energy to economic output will be in the strongest position.
“And some model providers may seem like they are making a lot of money because their models are very expensive… but that is short-term revenue growth,” Srinivas said.
Perplexity is increasing its focus on agency AI, a term that refers to AI systems that can perform more complex tasks beyond simple queries. In February, the company announced the Perplexity Computer, which it said can run complex tasks over long periods of time.
While Perplexity develops some of its own models, its core products integrate models from other AI companies such as Anthropic. One of the main focuses of Perplexity is to increase efficiency to achieve the best results while minimizing energy use. To support this goal, Perplexity on Tuesday announced Personal Computer, a tool it calls “orchestrator.”
The orchestration process involves a system that decides which model is best to use for a given task, how agents will work together, and where the AI should process queries. Today, most of the real AI operations are done in data centers.
But AI companies appear to be increasingly focusing on enabling these models to be rendered on a device such as a phone or laptop. Experts say this could reduce the power required to process AI, making it faster and more secure, since the data is not sent to a server. The Perplexity PC automatically directs processing to where it deems best.
“The data center comes to your laptop,” Srinivas said, adding that it is crucial to have an AI operating system that brings everything together in one unified system.
On Wednesday, Perplexity said the Personal Computer product will be available on: Microsoft’Windows operating system that allows AI to connect to applications such as Word and Outlook, as well as files on the user’s device. Perplexity has already launched the Personal Computer. AppleIt is a Mac product of .
Srinivas said Perplexity is focused on creating a “sustainable, lasting advantage” over competitors and that “it’s a regulatory issue.”
“We believe that by solving this we will build a resilient, highly valuable company with long-term advantage,” Srinivas said.
rising competition
The confusion faces increasing competition from rivals such as OpenAI, Anthropic and Google as their focus on AI agents increases.
Perplexity, which was last reported to be valued at $20 billion, comes on the heels of Anthropic and OpenAI, whose valuation rose to $20 billion. approximately $1 trillion and just over $850 billion, respectively. As investor demand for AI stocks continues, Anthropic confidentially filed for an initial public offering in the US this week.
Even as Perplexity expands its products into Microsoft and Apple products, the same companies are developing their own AI agents. On Tuesday, Microsoft announced new coding and reasoning models. Apple is currently working on an updated version of its digital assistant Siri based on Google’s artificial intelligence models.
Srinivas said Perplexity’s platform-agnostic approach will help these companies compete as they build their own AI.
“I think they will definitely try to build their own AI system, but we believe we’ve created the most versatile operating system by making it run on different models, different chips, different traditional operating systems, different hardware providers, different laptops,” Srinivas said.
“What we’ve done is this hybrid neutral regulation layer, and it allows us to balance all the different objectives at the same time.”
The CEO said that whenever Anthropic’s models improve, Perplexity also improves because Anthropic’s models are integrated into Perplexity. This means Perplexity has tripled its annual revenue since the beginning of the year “thanks to model advancements made by Anthropic,” he told CNBC.
