A Necessary Intervention: A Utilitarian Case for Governed SAI

We have run out of time for perfect solutions. The question now is whether we act with the imperfect ones. Global average temperatures are on track to exceed 2.7 to 3.1 degrees Celsius above pre-industrial levels by 2100 under current policy trajectories, a level of warming that would trigger irreversible tipping points across the cryosphere, monsoon systems, and global food supply.¹ Decarbonization remains the only permanent solution, but it is moving too slowly to prevent catastrophic overshoot. This essay argues that the deliberate, multilaterally governed deployment of stratospheric aerosols is a necessary complement to decarbonization, and that the serious objections to SAI demand better governance design rather than abandonment of the technology.

Inaction is not a neutral baseline

Opponents of SAI often argue from the precautionary principle: we should not intervene in a system as complex as the global atmosphere. In our current moment, however, inaction is not a neutral baseline. The atmosphere has already been intervened in, massively and recklessly, by two centuries of industrial emissions. The question is not whether humanity will alter the climate, but whether it will do so deliberately and accountably, or continue by default.

The evidence on comparative risk does not favor inaction. A 2022 review in Elementa: Science of the Anthropocene found that many of the health and ecological harms associated with SAI, including disrupted monsoons, altered UV exposure, and ozone changes, are largely comparable in magnitude to the harms already projected under unmitigated warming.² The risks critics attribute specifically to SAI are, in large part, risks that warming itself is already generating. What SAI adds is uncertainty about regional distribution; what unmitigated warming adds is certainty about catastrophic scale. A utilitarian calculus cannot treat inaction as the morally safe default, but it must also grapple honestly with which populations bear SAI’s uneven regional costs, a point this essay returns to below.

Decarbonization cannot win this race alone, but SAI cannot replace it

The structural obstacles to decarbonization are not failures of political will that better rhetoric will fix. They are the predictable outputs of a global capitalist economy in which the costs of transition are immediate and concentrated while the benefits are diffuse and delayed. Despite the Paris Agreement, global CO₂ emissions rose every year from 2015 to 2023 except during pandemic lockdowns. The International Energy Agency estimates that reaching net-zero requires $4 to $5 trillion in annual clean energy investment, roughly three times current levels.³ No enforcement mechanism exists to compel sovereign nations to meet these targets, and fossil fuel industries retain decisive lobbying power in democratic and autocratic states alike.

At an estimated $2.25 billion per year to offset half the rate of warming in its initial phase, SAI costs roughly one-thousandth of annual decarbonization expenditure.⁴ This cost advantage, however, carries a structural danger that must be named directly: termination shock. Because SAI requires continuous injection to maintain its cooling effect, stopping abruptly would release decades of masked warming in years rather than centuries. Once deployed, the catastrophic consequences of stopping create immense political pressure to maintain or expand the program indefinitely, potentially crowding out the decarbonization it was meant to enable. This is moral hazard not merely as an incentive problem but as structural lock-in. The necessary response is to build the constraint into the governance framework from the outset: deployment levels must be formally tied to verified, independently audited CO₂ emissions reductions, so that the program scales down as decarbonization scales up. Because termination shock makes rapid phase-out catastrophic, this scaling must operate on a gradual, decades-long timeline supervised by an independent scientific body with authority to adjust the pace. SAI without this mechanism is a bridge that quietly becomes a destination; SAI with it is an instrument of genuine transition.

Governance: enforcement without illusions

The most serious objection to SAI is political: who controls the global thermostat, and what stops a superpower from deploying in ways that protect its own agricultural belt at the expense of monsoon systems elsewhere? In 2022, a US startup conducted unauthorized aerosol releases in Mexico without international consequence, prompting Mexico to call for a geoengineering ban.⁵ Two consecutive UNEA sessions have failed to establish even a basic SAI scientific expert group, blocked by US and Saudi opposition; the UNEP’s own expert review concluded in 2023 that existing governance frameworks are wholly inadequate for the risks SAI presents.⁶ A 2025 Harvard International Law Journal analysis warns that unilateral SAI could be viewed as a national security threat by affected states, escalating into what the authors call “geoengineering wars”.⁵ Multilateral norms alone are insufficient: today’s superpowers have demonstrated they will exit agreements when national interests demand it.

The most credible deterrent is coordinated economic consequences rather than a military force or Security Council resolution, both subject to superpower veto. Any nation deploying SAI outside the agreed framework, or unilaterally adjusting deployment levels for national advantage, should face a pre-committed coalition of trade and financial sanctions activating automatically upon verified violation, without requiring a new political vote. The Montreal Protocol succeeded not through force but because exclusion from the trading coalition cost more than compliance. China needs export markets; the United States needs dollar-system cooperation. This mechanism has real limits. It assumes coalition cohesion under geopolitical strain, requires technically difficult and politically contestable attribution of violations, and risks retaliation that fractures the coalition. These weaknesses are genuine. But the relevant comparison is not a perfect institution; it is the current governance vacuum, which already enables unmonitored unilateral action with zero accountability.

Distributional justice is a further constraint the framework must address substantively. Peer-reviewed modeling shows that tropical injection strategies risk reducing precipitation specifically in monsoonal regions of South Asia, with effects highly dependent on injection latitude and quantity.⁷ Some regions may be net worse off under SAI even after accounting for warming avoided. A Loss and Damage fund, financed from the cost differential between SAI and equivalent decarbonization expenditure, is a necessary but insufficient response on its own. Affected regions must also have genuine procedural influence over deployment decisions, not merely compensation after the fact. Designing such consent mechanisms is genuinely difficult: structures that give affected nations veto rights risk paralyzing the program, while weaker arrangements risk being ignored. How to balance legitimacy against operability, how to guarantee the independence of the scientific review bodies that would administer both the phase-out timeline and attribution assessments, and how to prevent superpowers from appointing favorable experts to those bodies are unsolved governance problems. The argument here is not that they are easy to solve, but that they are tractable design challenges, not reasons to abandon the framework altogether.

Conclusion

The case for SAI is not that it is safe, politically simple, or free of distributional costs. Those who hold serious concerns about governance failure, lock-in risk, or procedural justice are raising legitimate objections, and this essay has tried to engage them honestly. The realistic alternatives – unmitigated warming or a governance vacuum that allows unilateral deployment – are worse by every measure. A well-governed SAI program, with a gradual phase-out tied to decarbonization, automatic sanctions, independent monitoring, and a Loss and Damage fund with scientific review, is a proportionate response to a crisis our existing tools are too slow to address.