Carbon dioxide capture and storage (CCS) technology is considered by many to be an essential route to meet climate mitigation targets in the power and industrial sectors. Deploying CCS technologies globally will first require a portfolio of large-scale demonstration projects. These first projects should assist learning by diversity, learning by replication, de-risking the technologies and developing viable business models. From 2005 to 2009, optimism about the pace of CCS rollout led to mutually independent efforts in the European Union, North America and Australia to assemble portfolios of projects. Since 2009, only a few of these many project proposals remain viable, but the initial rationales for demonstration have not been revisited in the face of changing circumstances. Here I argue that learning is now both more difficult and more important given the slow pace of deployment. Developing a more coordinated global portfolio will facilitate learning across projects and may determine whether CCS ever emerges from the demonstration phase.

Economic models deem rapid wide-scale deployment of CCS in the next few years to be essential in restraining the costs of meeting the 2 °C target for global temperature1,2, but CCS technologies are still at the pilot and demonstration phase. Paradoxically, it is primarily the costs of the early demonstration projects that have hampered further deployment. As each CCS ‘demonstration’ plant costs on the order of US$1 billion, during a time of fiscal austerity it has proved difficult to justify public support. Near-term pressure to develop CCS has also eased as most countries found it easier to meet their Kyoto targets because of the economic crisis (and other factors such as the US shale gas revolution). Meanwhile, unlocking private financing remains elusive and depends on developing necessary legal, institutional and commercial frameworks, as well as significant cost reductions and de-risking that can only come from operating multiple plants3.

Difficulties in justifying pilot and demonstration plants or deployment policy are hardly restricted to CCS, and can be found for nuclear power, renewables and indeed virtually any novel technology4,5, but the emphasis on demonstration is most common in the process industries6. At its broadest, CCS ‘demonstration’ has been identified as having a dozen or more manifestations, ranging from discourse creation to coalition formation7. I acknowledge the many important dimensions of demonstration, indeed, different disciplines have radically different conceptions of the nature of demonstration6. Given the overwhelming government and industry focus on cost reduction8,9, however, I use this as a test of how learning is operationalized. Governments should at least be able to construct a portfolio of projects along the dimension that they deem as central to the enterprise of demonstration.

The technical rationales for demonstrations being large-scale include understanding power system reliability and performance10 and adequately characterizing each geological formation11. As large-scale projects must store roughly 1 million tCO2 per year10,11, this scale requirement poses a number of challenges when seeking to learn from multiple projects.

In this Perspective, I explore the history of CCS demonstration in an effort to understand how the initial optimism about large-scale rollout led to multiple, uncoordinated efforts to learn from diversity. In the absence of widespread deployment of CCS, the projects that have endured do not form a coherent programme aimed at learning. Going forward, therefore, any effort to successfully re-launch CCS at scale will need to revisit the fundamental case for demonstration, including how best to derive the most learning from the billions of dollars already invested and that will need to be invested in the next wave of projects. There is a need for greater clarity over what time frame, at what scale, at what cost and to what end CCS demonstration is being pursued.