Fumblers Fund
People that don't fumble never have the ball.
Fumblers Fund invests in ideas that sound pretty dumb until you think about them for a long time. Specifically:
- Finding exploits in physics
- AI compilers
- Nuclear reactors – all sizes
- Anti-satellite satellites
- Open-source alphazero
- Lean for other disciplines like physics, chemistry and biology (+ the RL to solve them)
- Synthetic data generation
- Low-rank personality approximation
- Gradient descent 2.0
- Underwater SLAM
- Reverse-fragmenting data
- Drug-free brain programming
- Enabling innovations – all kinds
As a general principle, we do not invest in B2B SaaS. Mostly because we're allergic to money (see the name of the fund), but also because we believe that data should be unified. Keeping data on different vendors' servers is a bad idea, and we don't want to be a part of that (ask Larry why). If you're building B2B SaaS to end this practice, we'd love to talk to you.
One more thing: AI is going to be the biggest thing ever. This is easy to say because everything is bigger today, but we really mean it. That being said, we're mostly interested in enabling innovations using AI.
With every new technology people see the efficiency and sustaining innovations first: "We can do X cheaper and replace Y with chatbot." But those who profit the most from X being cheaper are the ones who are spending a lot on X, i.e. incumbents. Boring!
We're really interested in what new things AI enables: Text, code, imagery, audio, and 3D are now ~free to create. Isn't that awesome? Do something fun with that.
We review all memos and decks sent via electronic mail to fred@fumblersfund.com. If you are a limited partner, reach out to max@fumblersfund.com.