The bets, and the path
Hyuge's three bets, where it competes and where it doesn't, and the rules for how it takes capital.
Three bets
What we're betting on
Second on scale, first on understanding
We don't race to build the largest model, so that we can lead on how learning works and why systems behave as they do.
Money shaped to the mission
No investor or funder can buy control of what Hyuge commits to.
Promises with receipts
Every claim we make is published in a form an outsider can check.
The path
AI that learns from experience, tells you how sure it is, and keeps private information private.
Toward a science of learning that people and machines can both understand.
Between the two: research partnerships on real problems, each with a measure agreed in advance, and findings published with their confidence level. We describe the direction as a direction, not as a product.
Where we compete
Where we compete, and where we don't
Where we compete
- Problems where ideas, trust and documentation matter more than raw compute
- Organizations whose most valuable information is private
- Work where a result is only useful if you know how sure it is
Where we deliberately don't
- The frontier pretraining race
- General-purpose chat products
- Competing on model size
Capital
How we take capital
Only rules that are adopted today are listed here.
Our capital rules are being adopted. They'll appear here once they're in place.
Traction
Results
Updated as results arrive.