Analysing the relationship between energy consumption and economic growth is essential to achieve the goal of sustainable development. We employ hot spot analysis to discover the spatial agglomeration of GDP per capita and energy intensity in Guangdong, China, from 2005–2018. Furthermore, panel vector autoregression coupled with a system generalized method of moments is performed to examine the dynamic causal relationship between energy consumption and economic growth under the framework of the Cobb-Douglas production function. Using a multivariate model and grouped studies based on the differences in regional economic development, we show that the GDP per capita of the Pearl River Delta (PRD) is significantly higher than that of the peripheral municipalities. However, energy intensity shows an entirely different spatial distribution. The development of the regional economy depends on its own “assembling effect”. GDP explains approximately 68.3% of the total variation in energy consumption in the PRD and only approximately 34.5% of that in the peripheral municipalities. We do not confirm Granger causality between energy consumption and economic development. Guangdong can decrease its energy consumption growth without substantially sacrificing its economic growth. The analysis framework of this paper has significant implications for regions in balancing economic development and energy consumption.