The focus of this paper is twofold: firstly we make a case for the use of Computational Intelligence (CI) techniques in the modelling and/or prediction of global weather, CO2 emissions, climate change and similar endeavours. CI exploits processes found in Nature, albeit by way of software simulations on digital computers (i.e. in silico), and excel in particular at pattern recognition and/or classification. Moreover, they are characterized as being non-algorithmic, bottom-up, data-driven, and learn-by-example. The second focus of this paper is to propose the use of carbonrather than silicon-based computing, specifically in the form of DNA (or molecular) computing. Notwithstanding the unsolved difficulties with the latter (especially concerning Input/Output), its inherent massive parallelism has the potential to yield significant performance advantages. Finally, to come full circle, it could well eventuate that the inherent parallelism of DNA Computing could be brought to bear in the modelling/prediction endeavours mentioned previously.