Tipping points: where the evidence is firm, and where it is not

The mechanism, and its limits

The charter this site works to asks for a credible pathway from an event or trend to existential-scale harm. For climate, the pathway usually runs through one of two things. The first is a tipping point: a threshold beyond which a part of the Earth system shifts to a new state, often irreversibly on human timescales. The second is a correlated shock to the food system, where several major producing regions fail in the same year. Neither is a forecast of collapse. Each is a way that a chronic problem could become an acute one.

Coral reefs: the clearest case

The Global Tipping Points Report 2025, written by about 160 scientists and released in October 2025, concluded that warm-water coral reefs have already crossed their thermal tipping point and are experiencing unprecedented dieback. It put the reef threshold at about 1.2°C of warming, below the roughly 1.4°C now reached. The report's authors said that unless temperatures return to 1.2°C, and eventually lower, warm-water reefs will not be retained at any meaningful scale.¹

This is the most observable of the tipping points, because bleaching and dieback are being measured now. Two cautions apply. The 1.2°C threshold is the report's own synthesis. And the consequences, though severe for the people who depend on reefs for food, coastal protection and income, are regional. Coral loss is a serious harm; on its own it is not a route to civilisational collapse.

Ice sheets: committed, but slow

The same report describes the polar ice sheets as close to their thresholds, and says parts may have already crossed them. Crossing would commit the world to several metres of sea-level rise. The key word is commit: the rise would play out over centuries, not decades. Observed sea-level rise is already accelerating. The WMO reports a rate of 4.75 mm a year over 2012–2025, against 2.65 mm a year in 1993–2011.² A slow, multi-century commitment is a different kind of risk from a sudden one, but it is hard to reverse, and it affects the places where much of humanity lives.

The Atlantic circulation: the live argument

No tipping point has a wider spread of expert views. The Atlantic Meridional Overturning Circulation (AMOC) carries heat north through the Atlantic. Its weakening or collapse would disrupt weather mainly around the North Atlantic. Estimates differ sharply:

• A 2023 study in Nature Communications by Peter and Susanne Ditlevsen, using early-warning signals in observations, estimated collapse around 2057, within a range of 2025 to 2095 at 95% confidence, under current emissions. It is the most aggressive estimate and has been challenged.³
• A 2025 study in Environmental Research Letters by Drijfhout, van Westen and Rahmstorf, using CMIP6 climate-model runs extended beyond 2100, finds the deep overturning slowing drastically by 2100 and then shutting down in all high-emission scenarios, and in some intermediate and low-emission ones.³
• Two 2026 results moved risk estimates up: a report of research finding that the models predicting the largest slowdown are the most realistic, and an Open University preprint putting the chance that collapse is already locked in at 10–23%. The preprint is not peer reviewed, and the other result reaches us only through secondary reporting.³
• The IPCC's assessment is that a full collapse is unlikely this century, and some scientists still hold that view. In November 2025 Iceland's government designated a possible AMOC shutdown a national security threat.³

There is no consensus assessment. The honest summary is that the probability is uncertain, that recent work has pushed estimates up, and that the highest numbers come from unreviewed or model-extrapolated analysis.

Food: the shortest path to a societal shock

The most direct route from warming to large-scale human harm is food. Gaupp and colleagues found in a 2019 study in Nature Climate Change a rising risk that wheat, maize and soybean crops fail in several major breadbaskets at once, while for rice the risk has fallen. A follow-up study by the same group modelled warming of 1.5°C and 2°C. For maize, it found the risk of multiple-breadbasket failure rising from about 6% at present to about 40% at 1.5°C and 54% at 2°C, and concluded that risks increase disproportionately between the two levels.⁴

A June 2026 assessment by the Council on Strategic Risks and the Woodwell Climate Research Center projects, relative to 2010, roughly a doubling by 2040 of the risk that a given year's crop fails for Indian wheat and German maize, and more than a tripling of the small chance of a simultaneous wheat failure in India, France and Germany. It notes that adaptation options are limited and that some climate models may underestimate the risk.⁵ That body is a policy research organisation, so we treat the assessment as informed analysis, not as a peer-reviewed result.

What these studies do not show is what happens next. A simultaneous harvest failure is a plausible trigger for a food-price spike, export bans, hunger and political instability. How far that would cascade depends on stockpiles, trade rules and the response, none of which the crop models capture.

What would make it existential?

On the current evidence, no single tipping point is shown to threaten civilisational continuity by itself. The argument for concern is cumulative: several thresholds with long lag times, each hard to reverse, interacting with a food system that has little slack, in a political environment with weak coordination. That is a judgement about compounding risk and not a measured probability. It is reasonable to hold it and also reasonable to ask for the evidence behind each link in the chain. We will return to each link as new studies appear.