OpenAI is pursuing AI models capable of functioning as fully automated researchers, according to chief scientist Jakub Pachocki. "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 stated.
Pachocki projects a future where "you kind of have a whole research lab in a data center," as improvements in model capability enable extended autonomous operation without human intervention. The shift stems from capability gains that allow models to work longer periods independently, moving AI from tool to autonomous agent.
The vision requires containing powerful systems in isolated environments. Pachocki argues very powerful models should be "deployed in sandboxes cut off from anything they could break or use to cause harm." He acknowledged regulatory challenges: "I think this is a big challenge for governments to figure out."
The automated research push occurs alongside enterprise AI infrastructure buildout. Nvidia has partnered with Nebius to expand cloud AI infrastructure, while S&P Global acquired Enertel to strengthen AI capabilities. These deals reflect corporate technology spending shifts toward AI infrastructure and acquisitions.
Market reaction suggests investor caution about AI monetization speed. Major indices fell 1.4-1.6% despite aggressive infrastructure investment, occurring as the Federal Reserve maintains current interest rates. The gap between infrastructure spending and revenue realization creates valuation uncertainty for AI-focused companies.
The convergence of autonomous AI research capabilities and enterprise infrastructure investment marks a maturation phase for artificial intelligence deployment. Companies are building the computational foundation for AI systems while researchers push toward autonomous operation, concentrating technological capacity in fewer hands.
Corporate AI acquisitions target domain expertise rather than raw computing power. S&P Global's Enertel purchase exemplifies this pattern, adding specialized AI capabilities to existing platforms rather than building from scratch.
Sources:
1 MIT Technology Review, March 20, 2026
