AI-powered streetlights run on solar energy and generate revenue from AI services

Emerging as a new model in smart city infrastructure, iLamp aims to transform street lighting from a public utility that merely consumes electricity into a distributed artificial intelligence infrastructure. The solar-powered poles provide lighting at night without needing to be connected to the grid, while also offering inference services via the Nvidia-based edge AI hardware integrated into them.

Street lighting is a significant expense for municipalities. According to information provided at the source, this expenditure can account for 25 to 40 percent of total electricity consumption in some municipalities. The iLamp model aims to reduce this burden and even turn the poles into revenue-generating infrastructure units.

The technical structure of the system relies on storing energy produced during the day in a LiFePO4 battery. Depending on the model, the pole can generate between 200 and 600 watts of electricity, while consuming approximately 80 watts for nighttime lighting. When no movement or human presence is detected, the LED light level can be reduced to the 10-20 percent range, thereby achieving energy savings.

ON-SITE PROCESSING WITH EDGE AI

It is stated that the Nvidia Orin-based edge AI system inside the pole consumes approximately 25 watts. According to the startup's claims, this hardware can generate between 2,000 and 4,500 dollars in revenue per pole annually by providing inference services to clients similar to OpenAI. The fact that cooling needs are met by natural airflow within the pole is presented as one of the notable highlights compared to data center costs.

On the AI side, the primary area of use is the operation and inference of models rather than their training. For this reason, while Jetson-type edge AI devices are not sufficient for training large models, they can process image, sound, and sensor data in the field on-site.

This approach allows data to be interpreted on the pole instead of being constantly transferred to a central hub. For example, instead of sending the entire image from a traffic camera to a data center via a 5G line, the pole can transmit only processed information, such as the number of vehicles passing per second, to the center. This can reduce communication and data center costs while also limiting security risks.

It is reported that iLamp is not limited to just lighting and AI operations, but can be customized with more than 80 modules. Functions such as gunshot detection, license plate recognition, face and smoke detection, weather measurement, air pollution monitoring, electric vehicle charging, and drone charging stations are among these modules.

It is stated that the price of a pole capable of operating off-grid can reach up to 10,000 dollars. For this reason, it is expressed that the company prefers to obtain installation permits from municipalities and operate the infrastructure itself rather than selling the system directly. It is suggested that the model could pay for itself within a few years thanks to revenue generation.

The startup's scaling plans include the installation of 50,000 off-grid poles in Nigeria. It is stated that this installation will create a total of 13.75 petaOPS of AI processing power, and this capacity could be leased to large companies. In the security solution developed against theft, the goal is to render the processor unusable if an attempt is made to dismantle it.

If such systems become widespread, streetlights could take on a new role in cities, not only in terms of lighting but also in terms of data processing, sensor networking, communication, and energy management. However, issues such as security, privacy, cost, and public oversight will be decisive in how the technology is implemented.