Seminal
The magazine and atlas of EU-funded researchA new publication from the web and AI team at Scott Charnley

Atlas / United Kingdom

European Research Council grant · Running

THINKMETAL The Material Production of AI’s Cognition

Coordinator
University of Durham, United Kingdom
Runs
1 January 2027 to 31 December 2031
Status
Running
EU funding
€1.4m
Total cost
€1.4m
Partners
1 organisations in 1 countries
Type
European Research Council grant
Grant agreement
101304250

What the project is doing

In the project's own words, from its summary on CORDIS.

Until Thinking Metal, the attention of critical AI studies has overwhelmingly focused on software, including algorithms, code, parameters, and models. This left understudied how, where and by whom AI microchips and AI hardware are produced, how the materiality of microchips influences AI’s inferences and decisions, and how materiality allows us to conceive AI differently. Thinking Metal develops a “full-stack” approach to studying the material production of AI cognition, first by mapping the Global Production Network of AI microchips and their associated geopolitical dynamics and political economies.

Second, Thinking Metal unpacks how microchip architectures influence how AI “thinks” and acts in the world. Third, then, Thinking Metal will deploy this “full stack” approach to AI ethics and politics to understand how a focus on hardware can help us envision AI differently. Each objective of Thinking Metal addresses a gap in the existing literature: first, beyond a handful of well-known companies, there is a lack of research on the whole value chain and production network of AI microchips.

A complete appraisal of the political economy of AI microchip production is essential to understand how large corporations and hostile governments can capture, through infrastructural and hardware lock-in, current public and private research into AI algorithms. Second, despite the computer science literature showing that the same algorithm can remember and forget different features and data when deployed on different hardware configurations, critical data and algorithm studies have not included hardware in their investigations of AI ethics, fairness, bias, and sustainability.

Third, hardware will play a vital role in envisioning alternative forms and configurations of AI that can be more open and sustainable.

Topics

Who is involved

OrganisationCityCountryRole
University of DurhamDurhamUnited KingdomCoordinator

Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI