A Blueprint for Multinational Advanced AI Development

The multinational partnership should aim to minimize its use of compute located in non-member states and compute owned by foreign entities. Although it could begin by renting compute, this reinforces and creates substantial vulnerabilities over time....

guaranteed, cheap access to jointly-developed models would enable domestic firms to capture automation gains without dependency on foreign providers; structured licensing would stimulate local application ecosystems; and equitable benefit-sharing wou...

A semi-distributed structure, with a few core facilities, leveraging members' comparative advantages. Core facilities—fewer in number than the participating states—are necessary to create a critical mass of talent. Similarly, distributed compute can ...

Coordinated deployment of existing, planned, and within-reach European and other bridge power AI compute capacity is likely to provide sufficient computational resources to produce frontier AI models in the next few years, although significantly more...

Large-scale, international research and development collaborations such as CERN, Airbus, ITER, Intelsat, and the Human Genome Project demonstrate the strengths and weaknesses of different approaches. Some precedents of similar multilateral projects e...