According to online reports, OpenAI CEO Sam Altman predicted in his latest blog post that in 2025, agent systems capable of true cognitive work will emerge; in 2026, systems capable of proposing new insights may emerge; and in 2027, robots capable of performing tasks in the real world may appear. In some very important respects, the 2030s are likely to be very different from any period before, with intelligence and energy becoming extraordinarily rich, and with abundant intelligence and energy (and good management), one can theoretically have anything else. At the same time, we are also facing serious challenges. Security issues need to be addressed from a technical and social level, but considering the economic impact, widespread access to super intelligence is also crucial. Perhaps the best way forward is: 1. Solve the alignment problem, which can stably ensure that the learning and actions of the artificial intelligence system proceed in the direction that the collective truly wants in the long term;2. Focus on making superintelligence cheap, widely available, and not too concentrated on any individual, company, or country.