Previous technological revolutions often had slower competitive feedback loops. In AI, a breakthrough can become globally visible in hours, alter strategy in days, and be partially replicated by competitors within months, even though its full economic consequences may still take decades.
That could make technological leadership much more temporary. A company can look untouchable in January, get blindsided in March, change direction in April, and be competitive again by summer. Interesting times...
And yeah, some people won’t really see this because they don’t closely follow what the AI labs are shipping, or what’s happening across datacenters and inference providers. When a lab drops a new checkpoint, training trick, or piece of tooling, the race can start almost immediately, basically as soon as the announcement or technical details reach the right people.
People outside the bubble probably won’t know exactly what happened or when. They’ll mostly encounter the consequences later, as each month brings systems that are more capable and harder for institutions, companies, and society to keep up with.