Key insights:
- Perplexity Computer combines research, coding, design, and deployment in one system, reducing reliance on multiple AI tools.
- The company shifts to subscriptions and expands features, including patents search, shopping, and Galaxy voice assistant support.
- AMD and Meta sign a long-term AI infrastructure deal using Instinct GPUs and custom chips to power large-scale model training.
Perplexity introduces unified AI workspace
Perplexity AI unveiled Perplexity Computer, a platform that manages projects from idea to deployment inside a single environment. The system allows users to research information, design products, write code, and launch applications without switching services.
Introducing Perplexity Computer.
Computer unifies every current AI capability into one system.
It can research, design, code, deploy, and manage any project end-to-end. pic.twitter.com/dZUybl6VkY
— Perplexity (@perplexity_ai) February 25, 2026
The company reported that the platform also monitors live operations after deployment. Developers can review performance and adjust workflows directly within the interface. Perplexity aims to reduce fragmented workflows that often slow production across multiple AI tools.
Expanding products and subscription strategy
Perplexity has diversified its products over the last one year. In October 2025, it published Perplexity Patents, a component that enables end-users to query filings of global intellectual property using natural language queries. The tool is aimed at researchers, startups and law firms in need of quicker patent analysis.
The firm also partnered with Samsung to integrate its assistant, branded “Hey Plex,” into Galaxy devices. Meanwhile, U.S. Pro users gained an in-app shopping feature linked with commerce platforms such as Shopify. The company ended advertising trials and moved toward a subscription model, citing trust and answer neutrality as priorities.
AMD and Meta scale AI infrastructure
Separately, AMD (NASDAQ: AMD) and Meta (NASDAQ: META) announced a multiyear AI infrastructure agreement valued by analysts near $100 billion. AMD will supply up to six gigawatts of Instinct GPUs for training and inference workloads in Meta’s systems.
The companies will coordinate silicon, hardware systems, and software development to improve efficiency. Meta will also receive custom chips based on AMD’s MI450 architecture. Initial shipments are scheduled for the second half of the year.
Meta already operates millions of AMD EPYC processors and large numbers of MI300-series GPUs. The new agreement expands collaboration within the Open Compute Project and strengthens Meta’s computing capacity for future AI models.
