Perplexity launches Frontier Research unit to build an automated scientist
Perplexity is betting on machines that do science on their own. The company launched Perplexity Frontier Research, a unit backed by at least 10,000 GPUs and led by Andrew Gordon Wilson, to build an AI that runs the entire scientific process autonomously. It's also promising to publish its work openly.
- 2 sources report it: cryptobriefing.com, X account @Cointelegraph.
- The source published it on Nov 16, 2026 at 7:36 PM.
Nothing open. Every check passed.
Perplexity Frontier Research is now official. The new unit is chasing what the company calls an automated researcher — a system capable of running the entire scientific research process on its own. To get there, Perplexity has committed at least 10,000 GPUs to the effort.
The team is led by Andrew Gordon Wilson. Perplexity is recruiting for it across San Francisco, Palo Alto, New York and London.
The compute is earmarked for large-scale experiments in three core areas: pre-training, reinforcement learning and continual learning. The team's technical pillars run wider still, covering multi-agent orchestration, pre-training architectures, continual learning, RL, systems and specialized evaluations. Multi-agent orchestration is the standout — coordinating multiple AI agents, each handling a different function, and folding them into one research pipeline.
Perplexity has attached an open science commitment to the project, saying it will regularly publish papers, models and data.
The unit extends earlier Perplexity products, including Deep Research and Computer, which gave users AI-assisted ways to dig into questions and get tasks done. Perplexity is framing Frontier Research as work that could fundamentally change how scientific discovery is conducted.
TopicsPerplexity · Perplexity Frontier Research · Andrew Gordon Wilson · Deep Research · Computer · San Francisco · Palo Alto · New York
Written by AI from verified reporting.

