Upgrading hardware platforms and infrastruc- ture, such as personal computers, servers, or supercomputers, suffer from a tension when it comes to carbon neutrality. Increasing performances are an incentive for fast renewal, while manufacturing environmental costs urge us to extend as much as possible the lifetime of devices. Thus, carbon neutrality seems to be difficult to reach. In this article, we propose a theoretical model with an implementation available online. We study a refreshment of only the CPU and we try to answer quantitatively the question “what is the optimal point for which we may tend to carbon neutrality”? Our approach tackles a model of the greenhouse gas emissions, the primary energy and the monetary cost of two parameters that vary country by country: the use of a microprocessor, and its fabrication. We investigated three different renewal scenarios comparing two and three CPU versions: (1) maintaining a constant workload capacity, (2) maintaining the amount of processor cores fixed, and (3) keeping the amount of microprocessors fixed. Our results suggest that the most efficient solution (i.e with the lowest impacts) is to replace the entire batch of CPUs with the newer batch as soon as possible, which backs up the idea that modular replacement of hardware components is worth studying to help optimizing the impacts of computer equipment in general.