Energy efficiency is usually achieved by reducing the energy consumption as far as possible. With the growing amount of renewable energy sources, energy efficient usage also has to consider what kind of and when power is consumed. By matching the availability of electrical power with the current demands, the amount of unused energy and therefore overall energy production can be reduced. The so called smart grid aims to provide this matching with a broad deployment of smart meters to acquire the current demand. However, current approaches to the smart grid cannot handle the huge amount of sensors and energy sources involved in a scalable way. Most data acquisition systems focus on the lookup and reading of single sensors and therefore do not fit the requirements of a large scale power grid simulation. We propose a Global Sensor Grid (GSG) which provides consumers with data preprocessed to their needs instead of delivering raw sensor data. With this decoupling from the actual sensors, multiple consumers can benefit from improvements in data acquisition and avoidance of the redundant processing of data by each consumer. By integrating so-called diagnostic simulations into the GSG, gaps in sensor coverage can be filled with higher precision than normal interpolation.