From what I’ve seen, the usefulness of my AIs are proportional to the data I give them access to. The more data, (like health data, location data, bank data, calendar data, emails, social media feeds, browsing history, screen recordings, etc) - the more I can rely on them for.
On the enterprise side, businesses are interested in exploring AI for their huge data sets - but very hesitant to dump all their company IP across all their current systems into a single SaaS that, btw, is also providing AI services to their competitors.
Consumers are also getting uncomfortable with the current level of sharing personal data with SaaS vendors, becoming more aware of the risks of companies like Google and Facebook.
I just don’t see the winner-takes-all market happening for an AI powered 1984 telescreen in 2025.
The vibes I’m picking up from most everybody are:
1) Hardware and AI costs are going to shrink exponentially YoY
2) People do not want to dump their entire life and business into a single SaaS
All signs are pointing to local compute and on-prem seeing a resurgence.
I mean that was always how the route was going to go. Theres no way for them to recoup without either heavily on Saas, enterprise or embedded ads/marketing.
