New technologies demand new modes of governance to ensure they have beneficial impacts on societies. In this article, we develop a taxonomy of technology governance forms and consider the strengths and weaknesses of each. In particular, we consider the challenges and opportunities posed by advancing artificial intelligence. Subnational governance is prevalent and mitigates some risks, but it is insufficient when the individual rewards from taking risky actions outweigh normative sanctions. Nationally-enforced standards are promising ways to govern AI deployment, but they are less viable in “race-to-the-bottom” environments. When it comes to powerful technologies with military implications, a non-proliferation plus norms-of-use regime is the only multilateral option with a strong historical precedent. In plausible contexts, however, it is not effective on its own. Verification-based development- and use-restrictions would address a range of risks, but face challenges in the case of advanced AI and there is no clear example of major powers restricting the development of a powerful military technology that does not have a substitute technology. International monopoly, which might evolve naturally or be created, was the preferred solution of thinkers in the early nuclear era. It has the benefit of requiring less invasive monitoring, but at the possible cost of eroding national sovereignty. Overall, no single approach dominates the others. Recognizing these governance challenges facilitates the development of political and technical processes to overcome them.