Researchers at Switzerland's Federal Institute of Technology Zurich (ETH) have transferred approximately 100 petabytes of publicly available NASA climate data to a supercomputer for AI training. The initiative was announced late last week and has drawn attention due to its scale and timing. Reto Knutti, professor of climate physics and head of ETH's Center for Climate Systems Modeling (C2SM), described the data volume as 'impossible to visualise.'

The data transfer was initiated amid ongoing concerns about US funding for climate research. The United States has seen reductions in support for scientific initiatives, prompting institutions abroad to take greater responsibility for advancing climate modeling. The move by ETH highlights the growing role of international collaboration in addressing global environmental challenges.

The transfer of 100 petabytes of data is equivalent to the storage requirements of 20 million feature-length films. It took approximately a year to complete the process, involving the copying of roughly six billion files. Knutti noted that AI models trained on such data could significantly improve climate prediction accuracy and speed.

The implications of this data transfer extend beyond research. Increased reliance on foreign supercomputers for AI training may raise concerns about data sovereignty, governance, and long-term costs. The move could also influence how global institutions approach data sharing and model development in the future.

The situation remains under active discussion among researchers and policymakers. As AI continues to play a larger role in climate science, the balance between international collaboration and national interests will become increasingly important. The transfer underscores the need for clear frameworks to manage data usage and ensure equitable access to scientific advancements.