Time to build is a characteristic of most of new technology project mainly caused by physical or financial limitations. So, decision-makers have to take account the internal technical uncertainties of the project into when making investment choices instead of considering only external market conditions. This paper analyzes the implicit mechanism underpinning the new technology sequential investment decision in perspective of the patterns and efficiency of different learning in resolving or reducing different types of uncertainty. Especially, this paper introduces the learning heterogeneity in the resolution efficiency of technical uncertainty, which is compensatively determined by learning capability, accumulated and expanded learning effect. Adopting the real options approach, the mathematical model is constructed and solved by numerical method, with comparative static analysis conducted as well. According to this, I expect to give more insight into the investment mechanism of and provide rational decision-making guideline for new technology projects.