CMIP7 Scenarios: What Changed, What It Means, and What Comes Next for Climate Risk

Written By Santiago Ponce

On 7 April 2026, the Scenario Model Intercomparison Project published the experimental design for CMIP7 in Geoscientific Model Development, the peer-reviewed journal used by the climate modeling community to document its protocols. On 20 May, the IPCC issued a public statement clarifying its relationship with the new scenarios. Both developments have generated significant commentary, some of it careful, some of it not.

This post explains what CMIP7 scenarios are, what has changed compared to the previous generation, what the IPCC has and has not said, and what this means for organizations that use climate scenarios in risk assessments, disclosure, or strategic planning.

What CMIP7 scenarios are

The Coupled Model Intercomparison Project (CMIP) is coordinated by the World Climate Research Program (WCRP). It organizes standardized experiments that Earth system models run using common inputs, so their results can be compared, aggregated, and used in scientific assessments. CMIP scenarios provide the emissions and land-use pathways that drive those models. The results of previous CMIP rounds have formed the quantitative backbone of every recent IPCC assessment report.

ScenarioMIP, the component of CMIP responsible for scenario design, has proposed seven scenarios for the CMIP7 round. They are labeled by their emissions trajectory, not their temperature outcome (which will be determined by the model runs themselves):

  • High (H): Emissions as high as judged plausible, assuming policy rollback and continued fossil fuel dependence. Expected warming of approximately 3.0°C by 2100.

  • High-to-Low (HL): Follows the High pathway until mid-century, then applies deep mitigation to reach net-zero CO₂ by 2100.

  • Medium (M): A continuation of current implemented policies, without additional ambition. Not a "most likely" scenario, but a benchmark for existing policy levels.

  • Medium-to-Low (ML): Delayed strengthening of mitigation, reaching net-zero CO₂ by the end of the century, with net-negative emissions thereafter.

  • Low (L): Consistent with staying likely below 2°C, with rapid near-term reductions aligned with current NDC pledges.

  • Very Low (VL): Designed to keep warming as close to 1.5°C as judged feasible, with limited overshoot.

  • Low-to-Negative (LN): Higher overshoot of 1.5°C followed by large-scale carbon dioxide removal to return temperatures to lower levels.

All scenarios cover the period to 2100, with stylized extensions to 2500 to support research on long-term dynamics, irreversibility, and tipping points. A significant innovation is that most scenarios will be run in emission-driven mode for CO₂, meaning the Earth system models calculate atmospheric concentrations internally rather than having them prescribed. This better captures carbon cycle uncertainties but will produce a wider range of temperature outcomes across models.

What changed from CMIP6

The most visible change is the absence of SSP5-8.5, the extremely high-emissions scenario that has been widely used in climate risk analysis, infrastructure planning, and regulatory stress tests for over a decade. SSP5-8.5 assumed a world of massive coal expansion (five to six times current levels by 2100), a global population of approximately 12 billion, and minimal technological progress in clean energy. The ScenarioMIP authors judge this pathway as no longer plausible, citing trends in renewable energy costs, the emergence of climate policy across major economies, and recent emissions trajectories.

The new High scenario (H) peaks at roughly 71 GtCO₂/year in 2100, compared to 128 Gt under SSP5-8.5. Expected warming is approximately 3.0°C, compared to approximately 4.4°C. This is a meaningful downward revision of the upper boundary of the core scenario set.

At the low end, scenarios included in CMIP6 that have since become inconsistent with observed trends during the 2020 to 2030 period have also been revised. The overall range of the new set, from approximately 1.5°C to approximately 3.5°C by 2100, is narrower than CMIP6.

Other notable design changes include:

  • Overshoot scenarios are now a central feature, with multiple pathways exploring what happens when temperatures temporarily exceed a target before declining through net-negative emissions.

  • Extensions to 2500 allow research into long-term system behavior. Under the High extension, temperatures could reach approximately 6°C above preindustrial levels. Several LinkedIn commentators have noted that treating 2100 as a finish line is misleading: under higher-emissions futures, temperatures may not reach 4°C by 2100 but may continue rising well beyond it.

  • Carbon dioxide removal (CDR) is explicitly integrated across scenarios, with detailed protocols for how afforestation, BECCS, direct air capture, and enhanced weathering are represented in models.

  • Plausibility criteria are applied systematically. The authors acknowledge that plausibility is subjective and evolves, but they are explicit that the High and Very Low scenarios push the boundaries of what is considered plausible by design, and are therefore expected to be considered less likely than those in the middle of the range.

What the IPCC has said

On 20 May 2026, the IPCC issued a public statement in response to media and social media reporting that had attributed the CMIP7 scenarios to the Panel. The statement is unambiguous:

  • The ScenarioMIP paper is not from the IPCC.

  • The IPCC does not conduct its own research, run models, or own any scenarios.

  • The CMIP7 scenarios are part of the broader scientific literature coordinated by WCRP.

  • They fall within the scope of the IPCC's next assessment (AR7) and will be considered during that process.

  • The IPCC does not comment on individual papers.

This distinction matters. The CMIP7 scenarios have been endorsed by the WCRP's Joint Scientific Committee (in February 2025), which described them as a "scientifically rigorous framework for assessing future climate pathways." But WCRP endorsement is not the same as IPCC endorsement. The IPCC assesses the full body of peer-reviewed literature. The CMIP7 scenarios will be one of many inputs in the AR7 cycle.

For organizations that reference "IPCC scenarios" in their climate risk disclosures or strategy documents, this is worth noting. The scenarios assessed in AR6 (based on CMIP6) remain the current IPCC-assessed set. CMIP7 scenarios will not carry that status until the AR7 reports are published, with the Synthesis Report expected by late 2029.

What this means for climate risk practice

Several practical implications follow from CMIP7, but they unfold over different time horizons.

Near term (2026 to 2027): Earth system model simulations using CMIP7 scenarios are scheduled to begin in mid-2026, with some Tier 1 results expected by late 2026 or early 2027. However, usable, spatially resolved climate data at the resolution needed for corporate risk assessments, adaptation planning, or infrastructure design will take longer to produce. Downscaling, bias correction, and impact modeling all follow the raw ESM outputs. Existing CMIP6-based analyses remain valid and are still the standard reference for regulatory frameworks, including TCFD, ISSB, and CSRD climate risk disclosures.

Medium term (2027 to 2029): As CMIP7 data becomes available and peer-reviewed literature incorporating the new scenarios begins to accumulate, climate risk service providers, scenario platforms, and regulatory guidance documents will begin to update. Companies that use third-party climate risk tools should expect their providers to migrate to CMIP7 data during this period. The EPRI technical update published in March 2026 already encourages companies to understand newer emissions pathways and their implications for target-setting and transition risk assessment.

Longer term (2029 and beyond): The AR7 Synthesis Report, expected by late 2029, will provide the assessed scientific basis that regulatory frameworks and disclosure standards are likely to reference. At that point, CMIP7 scenarios will replace CMIP6 as the standard reference for climate projections in corporate and policy contexts.

Three points that deserve more careful attention

Commentary on LinkedIn and elsewhere has highlighted several important nuances worth examining more closely.

Until approximately 2050, scenario differences remain small. The warming trajectory for the next two decades is largely determined by emissions already in the atmosphere. For risk assessments with a 20- to 30-year horizon, particularly those focused on physical risks to assets and infrastructure, existing analyses based on CMIP6 remain robust. The divergence between scenarios becomes significant primarily in the second half of the century.

The retirement of SSP5-8.5 does not mean high warming is impossible. SSP5-8.5 described a specific emissions pathway that is now considered implausible. But similar or higher warming levels could still emerge from different combinations of factors: higher climate sensitivity than assumed in central estimates, stronger carbon cycle feedbacks (for example, permafrost thawing or reduced ocean carbon uptake), or cascading system failures. The physical risk associated with high warming has not been revised downward. What has changed is the assessed plausibility of one specific pathway to reach it.

Climate sensitivity estimates may be higher than previously assumed. Several commentators have noted that while the emissions outlook has improved, estimates of climate sensitivity, the degree to which the climate system responds to a given concentration of greenhouse gases, appear to be trending upward. If confirmed by CMIP7 model runs, this would mean that even moderate emissions pathways could produce more severe warming and impact outcomes than earlier models suggested. This reinforces the importance of not interpreting the narrower scenario range as a narrower risk range.

What to watch

For organizations using climate scenarios in reporting, risk assessment, or strategic planning, the key developments to monitor are:

  • Mid-2026: First CMIP7 Earth system model outputs become available. These will be raw, global-scale results, not yet suitable for corporate-level physical risk analysis.

  • 2027 to 2028: Climate risk data providers and scenario platforms begin incorporating CMIP7 data. Downscaled, bias-corrected products start to appear.

  • 2027: IPCC Special Report on Climate Change and Cities, which may be among the first IPCC products to reference CMIP7 literature.

  • Late 2029: AR7 Synthesis Report provides the formally assessed scientific basis that disclosure frameworks are likely to adopt.

In the interim, companies do not need to revise their climate risk assessments or scenario analyses based solely on CMIP7. What they should do is understand the direction of travel: a scenario architecture that is more refined, more explicit about plausibility, and more demanding in how uncertainty is represented. The tools are changing. The underlying physics are not.

Sources

Santiago Ponce
Previous

Ransomware Gets a Playbook. El Niño 2026 Gets a Footnote.
Next

Voluntary Standard, Mandatory Consequences: What the EU's New Sustainability Reporting Framework Means for Smaller Businesses