AI2 Incubator, which emerged from the Allen Institute for AI in 2022, has obtained a substantial $200 million in computing resources that startups in its program can utilize to speed up their initial growth.
“Our community of hundreds of AI practitioners is in great need of compute,” stated managing director Jacob Colker. Many startups struggle to demonstrate early success due to limited resources for advanced model training beyond standard API offerings.
Any company within the AI2 Incubator portfolio or program is eligible to receive up to $1 million worth of dedicated AI-style compute at data centers owned by an undisclosed partner. Colker declined to identify. However, the number of companies with $200 million of capacity available is quite limited.
The partner does not receive any special privileges or exclusive access to the companies, except for being the primary compute provider that a startup typically uses first.
Colker explained that there is a strong desire to help entrepreneurs generate revenue quickly.
A million dollars of dedicated compute can greatly benefit pre-seed startups, which is the main focus of AI2 (they have supported WellSaid Labs, Xnor.ai, and other companies through their program).
Colker indicated that it would meet the majority of computational requirements, even for businesses working on creating new foundational models.
“Various companies have varying requirements, but it’s not just about cloud credits; we offer dedicated machines and customized silicon.” This represents the largest computer allocation currently accessible to startups.
The AI2 Incubator has been in operation since 2017, but it became independent in 2022. It has a close relationship with the Seattle Research Institute but is now formally separate from it. They have contributed to the development of over 30 startups and secured a $30 million fund last year to sustain their efforts.
