Perplexity AI CEO Aravind Srinivas has outlined a key strategy for winning the artificial intelligence race, emphasising that the company which generates the most economic value from the energy consumed by AI systems will emerge as the long-term leader. In an interview with CNBC, Srinivas introduced the metric 'token value per watt per user' as the most important measure for AI companies, combining efficiency, performance and economic output.
Token Value per Watt per User
According to Srinivas, maximising this metric is more critical than simply building larger or more expensive AI models. 'Whoever is able to maximise this particular objective really will, by balancing accuracy, latency, cost, privacy and intelligence altogether; they're going to win – that's what's going to win long term,' he told CNBC. A token represents the basic unit of data processed by an AI model, with each prompt broken down into tokens that require computing power and energy. Srinivas argued that companies delivering the highest value from energy usage will be in the strongest position over time.
He also cautioned that high-priced AI models generating strong revenue today may not guarantee long-term leadership. 'And so it might feel like some model providers are making a lot of money because their models are very expensive ... but that's short-term revenue growth,' Srinivas added.
Perplexity's Focus on Efficiency
Srinivas's comments come as Perplexity expands its efforts in agentic AI, which refers to systems capable of carrying out complex tasks with minimal user intervention. Earlier this year, the company introduced Perplexity Computer, an AI agent designed for extended task handling. To improve efficiency, Perplexity recently launched Personal Computer, a product described as an 'orchestrator' that determines which AI model to use for a task, how agents collaborate, and where processing occurs.
Srinivas envisions a future where more processing moves from large data centres to local devices like laptops and smartphones. 'The data center is coming to your laptop,' he said, adding that AI systems require a unified platform to coordinate models, hardware, and operating systems. Perplexity has expanded the Personal Computer platform to Microsoft's Windows, enabling interaction with apps like Word and Outlook, as well as local files. The product was previously available on Apple's Mac platform.
Competition and Platform Neutrality
Perplexity faces competition from OpenAI, Anthropic, and Google, all investing heavily in AI agents and foundation models. Srinivas emphasised Perplexity's platform-neutral approach, supporting multiple AI models, hardware platforms, and operating systems. 'I think they 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,' he said.
According to Srinivas, improvements by model providers like Anthropic directly benefit Perplexity, as their technologies are integrated into the company's products. These advancements have helped Perplexity triple its annualised revenue since the start of the year.
