OpenAI Chief Scientist Jakub Pachocki predicts AI models will soon enable "a whole research lab in a data center," with systems conducting research autonomously for extended periods—a capability that could reshape how scientific breakthroughs emerge globally.1
"I think we are getting close to a point where we'll have models capable of working indefinitely in a coherent way just like people do," Pachocki said.1 These systems would operate longer without human intervention, moving beyond current limitations on autonomous task completion.1
The infrastructure requirements are substantial. Pachocki advocates deploying the most powerful models in isolated sandboxes, disconnected from systems they could damage or exploit.1 This approach reflects growing international concern about AI systems with extended operational autonomy, particularly as different nations develop competing governance frameworks.
The prediction arrives as AI infrastructure investment accelerates globally. Early 2026 saw 68 companies reach unicorn status in the sector,2 while established players make strategic moves—S&P Global acquired energy AI firm Enertel AI to strengthen analytics capabilities. The venture capital landscape shows concentration among specialized investors backing this new unicorn class.2
Pachocki acknowledges the governance gap. "I think this is a big challenge for governments to figure out," he said, recognizing that industry self-regulation proves insufficient for systems operating at research-lab scale.1 This challenge extends across jurisdictions, with the U.S., EU, China, and other powers developing divergent regulatory approaches to advanced AI systems.
Automated research capabilities could compress development timelines across AI labs worldwide, potentially accelerating breakthroughs while concentrating research capacity among organizations with sufficient computational resources. The data center model favors players with infrastructure scale, raising questions about global access to cutting-edge AI research capabilities and whether smaller nations and institutions can compete.
The regulatory frameworks Pachocki identifies as necessary remain undeveloped internationally, creating a gap between technical capability and governance readiness that affects researchers, policymakers, and technology companies across borders.
Sources:
1 MIT Technology Review, March 20, 2026
2 Crunchbase News, March 17, 2026
