Home / Technology / UK Startup to Embed AI Compute in Lampposts
UK Startup to Embed AI Compute in Lampposts
15 May
Summary
- Micro data centers could be embedded within urban street furniture like lampposts.
- Self-destructing chips will be used for enhanced data privacy and security.
- The plan aims to reduce latency and improve sustainability of AI compute.

UK startup Conflow Power Group Limited (CPG) is pioneering a new concept for data centers, aiming to decentralize compute power by embedding micro data centers within urban street infrastructure, notably lampposts. This innovative approach seeks to address major criticisms of current hyperscale facilities, particularly their environmental impact and energy consumption. By dispersing compute and integrating local solar energy generation with battery backup systems, CPG intends to create more sustainable data centers.
The initiative aims to significantly reduce latency and alleviate strain on national telecommunications networks by bringing AI compute closer to end-users and devices. CPG proposes utilizing more affordable Nvidia AI accelerators, costing around $2,000, as an alternative to high-end GPUs. A key security feature includes self-destructing chips with firmware locking and encryption to protect sensitive information if hardware is compromised or accessed unauthorizedly.
This hyperlocalization of data centers offers numerous benefits, including cost savings on data transfers and improved energy efficiency. Unlike massive data centers that can rival small towns in energy consumption, spreading out power usage geographically addresses grid connection delays and supply constraints. Lampposts are ideal due to their existing electrical and fiber optic connections, dense urban distribution, and existing use for other functions like 5G cells and EV chargers.
Furthermore, this distributed model aligns with growing geopolitical concerns regarding sovereign compute capabilities, particularly in Europe and the UK. It supports national drives for increased local data processing. However, CPG must overcome challenges such as weatherproofing, vandalism protection, and effective thermal management for these embedded units to become a widespread reality.