When I first started learning about AI, I assumed the model would always be the most valuable part.
The better the model, the stronger the business.
Lately I am not so sure.
Open-source models keep improving.
New models appear almost every week.
Over time, model quality may become less of a differentiator than people expect.
If that happens, where does the value move?
While reading about @OpenGradient , I started wondering whether infrastructure becomes more important as models become easier to access.
Anyone can download a model.
Not everyone can provide reliable inference, verification, coordination, and trust at scale.
Maybe the long term competition is not model vs model.
Maybe it's infrastructure vs infrastructure.
The companies building the roads may end up capturing more value than the vehicles driving on them.
That's one reason OpenGradient feels interesting to me.
It is not trying to win the model race.
It's trying to build the layer that remains useful no matter which model wins.