Perplexity CEO says ‘Orchestration’ is Crucial for AI Firms

Perplexity CEO Aravind Srinivas says companies that can provide the most economic value from the power their AI uses will ultimately receive the highest valuations.

He adds that any company that can provide the “most taken value per watt per user” will be the most successful.

“Whoever is able to maximise this particular objective really will, by balancing accuracy, latency, cost, privacy and intelligence all together, they’re going to win long term,” Aravind told CNBC on Wednesday.

When an AI chatbot is asked to carry out a task, it breaks it down into tokens – a basic unit of data that an AI model can process. Each token then requires energy to be processed. 

Aravind believes that companies which can provide the best balance of energy to economic output will ultimately be in the strongest position.

Integrating Anthropic models

This year, Perplexity has been increasing its agentic AI strategy.

In February, the company introduced Perplexity Computer, an agent with the ability to execute complex, multi-step workflow tasks – such as deep research, write reports and integrate with workplace tools – capable of running over a period of hours or months.

While Perplexity develops its own agents, its key products integrate models from larger AI firms like Anthropic.

One of the company’s key strategies is to improve efficiency to achieve the best outcomes while minimising energy use.

To support that goal, Perplexity recently announced Personal Computer, a tool which it calls an “orchestrator”.

On Wednesday, Perplexity said Personal Computer will be available on Microsoft’s Windows operating system, allowing the model to connect to apps like Word and Outlook, in addition to files on a user’s device.

Perplexity has already launched the Personal Computer model on Apple’s Mac product.

AI firms are increasingly looking to integrate these models within software but also across different hardware, such as phones or laptops.

Tech industry experts say this could reduce the power required to process AI as well as improving its speed and security, since the data is not being sent to a server.

Perplexity Personal Computer automatically directs the processing via the most efficient route.

“The data centre is coming to your laptop,” Aravind says, emphasising the importance of having an AI operating system that brings everything together in a single unified system.

“We believe that by solving that, we’ll be building a pretty valuable company that has endurable, long-term advantage.”

A platform-agnostic approach

Amid the development of its own AI models, Perplexity is facing increasing competition as rivals like OpenAI, Anthropic and Google have accelerated their focus on agentic AI.

Perplexity’s valuation was last reported at US$20bn, far behind Anthropic’s and OpenAI’s, who have been valued to nearly US$1tn and more than US$850bn, respectively.

Aravind says that while these companies – and larger tech companies like Microsoft and Apple – will build their own AI, Perplexity’s platform-agnostic approach will help it compete.

He says: “I think [companies will] absolutely will try to build their own AI systems, but we believe we’re building the most versatile operating system by making it work across different models, across different chips, across different traditional operating systems, different hardware providers, different laptops.”

This integration of Anthropic models has resulted in Perplexity tripling its annualised revenue since the beginning of 2026, “thanks to model advances that have been made by Anthropic”.