Smart cities generate data at the edge. The intelligence has to live there too.
The defining challenge of smart city AI is not processing power. It is power. Edge nodes — the compute units embedded in urban infrastructure, autonomous vehicle networks, environmental monitoring arrays, and public safety systems — must process high-volume, real-time data streams with strict latency requirements. They must do so continuously, in the field, often without reliable access to centralised power or connectivity.
Binary GPU accelerators are too power-hungry for most edge deployments. CPU-based inference is too slow. The result is a forced compromise between intelligence and deployability that has constrained smart city AI for a decade.
At 180W with 96% lower power generation than equivalent binary inference hardware, the Aterna G1 card changes the economics of edge AI. Compute that previously required dedicated cooling infrastructure and multi-kilowatt power budgets can now operate within the power envelope of standard industrial enclosures.
Aterna is engaged with partners developing AI infrastructure for next-generation urban environments, including projects aligned with Saudi Vision 2030 smart city initiatives. Edge Intelligence deployments under evaluation include:
Every Aterna edge deployment maintains the same sovereignty properties as our data centre hardware: no foreign cloud dependency, full data residency within the deployment jurisdiction, and post-quantum resilient communications for node-to-node and node-to-centre data exchange.
Fill out the form to request the document