Skip to main content
European data
data.europa.eu
The official portal for European data

Guidelines for developing an AI computing capacity plan

Learn about the key highlights of the report on guidelines for building national computing capacity for Artificial Intelligence (AI)

Artificial Intelligence (AI) has the potential to transform economies and offers opportunities for productivity, growth, and resilience. In response to this, many countries develop national AI strategies. However, there seems to be a lack of data or a targeted plan for national AI compute capacity required for AI systems to perform tasks. To help policy-makers assess and plan national AI compute capacity, the OECD.AI wrote a report ‘A blueprint for building national compute capacity for artificial intelligence’.

The report offers guidance on how to develop a national AI computing plan. There are three dimensions identified: capacity, effectiveness, and resilience. Capacity is about the availability of the right tools to ensure that investments reflect needs and consider the fast-changing pace of AI. Therefore, it is important to gain insights into the current and future demands of AI computing. The second dimension looks at effectiveness in terms of access to the resources needed, such as skilled people, policy systems and finances. Resilience relates to the security, sustainability and sovereignty of AI, such as the location of the computing capacity and responsible parties.

In addition to the guidelines for developing an AI computing plan, the report also offers indicators, datasets and proxies to measure national AI computing capacity and identifies gaps to these measurement tools, for example a lack of workers with adequate skills to access AI computing.

Artificial Intelligence offers many opportunities if the right computing capacity and strategy are in place. Read the full report for guidance on the development of an AI computing capacity and related strategy.

For more news and events, follow us on Twitter, Facebook and LinkedIn, or subscribe to our newsletter.

Text of this article