OpenAI's Brockman sees AGI in years

The important part is less the exact timeline than the posture: OpenAI is talking about AGI as a near-term operating assumption, which changes how people should read its product and infrastructure choices.

• –If AGI is “a couple years” away, the bottleneck shifts from raw model demos to reliability, long-context memory, tool use, and evals that actually survive production
• –The quote reinforces the compute arms race: near-term AGI beliefs justify massive capex, tighter NVIDIA/OpenAI alignment, and aggressive scaling plans
• –For developers, this is a reminder that “AGI” rhetoric is still mostly a strategic narrative, not a spec; the practical question is which capabilities become dependable enough to ship
• –The clip adds to a widening timeline spread across the industry, where leaders publicly disagree on whether AGI is already here, years away, or a decade out
• –Expect more roadmap decisions to be framed around “agent readiness” and “superintelligence prep” rather than incremental chatbot improvements