Keep pulling the thread on Demis Hassabis.
In 2016, a system created by DeepMind surpassed the world's best human players at the game of Go.
In 2020, DeepMind solved the protein folding problem by creating a system that could predict the shape of nearly all proteins in nature.
In 2024, Demis Hassabis and John Jumper were awarded the Nobel Prize in Chemistry for their work on protein structure prediction.
A primary motivation for Demis Hassabis to sell DeepMind to Google was to escape the continuous and distracting process of fundraising from venture capital investors.
Founders Fund's initial investment in DeepMind was $2.3 million for just under 50% of the company.
Larry Page successfully recruited Demis Hassabis to sell DeepMind to Google by arguing that using Google's existing resources would be a faster path to achieving AGI.
Elon Musk attempted to acquire DeepMind with either Tesla or SpaceX to prevent the Google acquisition, but Demis Hassabis rejected the offers due to insufficient cash flow from Tesla and inadequate computing power at SpaceX.
Google acquired DeepMind in January 2014 for $650 million.
Shortly after being acquired by Google, DeepMind's annual staff costs reached $260 million, six times its total spending in its first three years as an independent company.
Post-acquisition, a single research team at DeepMind would sometimes use more computational resources in a week than the entire global Gmail network, which served 900 million users at the time.
Sam Altman proposed the creation of OpenAI to Elon Musk with the rationale that since AGI development is inevitable, an entity other than Google should be the first to achieve it.
The AlphaGo Zero system was designed to learn the game of Go exclusively through self-play, without being trained on human expert games.