A new study says how fast we warm, not just how much, may decide whether the Atlantic's overturning circulation collapses

The Atlantic Ocean's big overturning current is one of the most-watched tipping elements in the climate system, and most of the public conversation about it has fixated on a number: how many degrees of warming before it collapses. A new modelling study argues that the question itself may be framed wrong, and that the speed of warming could matter as much as the amount.

What the study found

The paper, "Failure to track a stable AMOC state under rapid climate change," comes from René van Westen, Reyk Börner and Henk Dijkstra at the Institute for Marine and Atmospheric Research at Utrecht University, and was published in Nature Climate Change on 13 August 2026 (an open preprint has been available since February).

Their central result is that in their simulations, AMOC collapse was not set by a fixed temperature threshold but by the rate at which the climate was forced. Run the same model with different speeds of carbon-dioxide increase and it behaved very differently:

• A slow increase (about 0.5 ppm of CO2 per year) left the current stable up to about 5.5C of global warming (the end of that model run).
• A fast increase (about 2.5 ppm per year, close to the observed rate of recent years) tipped the current into collapse at only about 2C.
• An even faster increase (about 5 ppm per year) collapsed it at about the same low temperature, roughly 2C.

In other words, two futures that end up at the same temperature can end up on opposite sides of a tipping point, depending on how quickly they got there. Van Westen put it directly: "Our results show there is not necessarily a fixed temperature beyond which the AMOC inevitably collapses."

Why speed would matter

The proposed mechanism is about whether the ocean can keep up. As Dijkstra put it, "Under slow warming, the entire ocean, from the surface down to its deepest layers, has time to gradually reorganize and adapt to the changing conditions." In the model, slow forcing lets stabilising processes take hold: increased evaporation over the Atlantic makes surface water saltier and denser (which helps the overturning keep going), and shrinking sea ice changes how much fresh meltwater enters the system. Given enough time, those effects counteract the freshening that would otherwise shut the current down.

In the model, push the system too fast and it cannot track that moving target. The stabilising adjustments happen on a timescale slower than rapid forcing, so under fast warming the destabilising feedbacks win and the current tips. The team used a full coupled climate model (a version of the Community Earth System Model) for the main experiments and a simple conceptual "box" model to isolate and explain the rate-dependent mechanism.

This reframes a decade of debate. Earlier estimates put an AMOC tipping point somewhere around 4C of warming, with a wide range, and treated it largely as a temperature threshold. This study's argument is that the threshold is not really a temperature at all; it is a race between how fast the climate is pushed and how fast the ocean can adapt.

The caveats the headlines will drop

This is where discipline is needed, because the result is easy to misread as "the ocean can handle over 5C, so relax." The authors go out of their way to say the opposite.

First, this is one model. The main experiments use a single climate model plus a deliberately simplified box model that, in the authors' own words, is a strong simplification capturing only one rate-dependent mechanism. Different models can behave differently, the salinity signals that drive AMOC stability are small and noisy, and although most current models line up qualitatively with this one, the details vary. A finding from one model is a hypothesis about the real world, not a measurement of it.

Second, and most important, the team is explicit that their result is not reassurance. Their model only tips when Greenland's meltwater input is pushed to something like 65 times the present-day rate, a threshold they attribute to known biases that make climate models too stable. Because of that, they say, the real AMOC is likely closer to a tipping point than their simulation is. The "still stable at 5.5C under slow warming" number is a property of the model's stability, not a safety margin for the planet.

Third, the scenarios are idealised. The experiments use smooth, linear increases in CO2 rather than realistic emissions pathways, which is a standard way to probe a mechanism cleanly but is not a forecast of any particular future. And the study is about the conditions for collapse in a model, not observational evidence that the real current is or is not about to tip.

Why it matters anyway

Even hedged, the shift in framing is significant. If the rate of change, not just the amount, helps decide whether a major tipping element goes over the edge, then the speed of decarbonisation carries weight beyond its effect on peak temperature: warming slowly to a given level could be meaningfully safer than warming quickly to the same level. The study offers a rough equivalent, a critical warming rate on the order of 0.3C per decade, and notes the recent observed pace of CO2 rise sits in the range that tipped the current in the model.

The honest bottom line is two-sided. The mechanism, that slow change can let a system adapt while fast change overwhelms it, is intuitive and, if it holds across other models, genuinely useful for thinking about tipping points. But it comes from a single idealised study whose own authors warn the real ocean is probably more fragile than their simulation, and it is not evidence that the AMOC is safe at any particular temperature. The next step is whether other modelling groups reproduce the rate dependence. Until then it is a compelling idea, not a settled fact.

The study at a glance

Published	Around 14 August 2026, Nature Climate Change
Who	Van Westen, Börner and Dijkstra, Utrecht University
Question	What triggers a collapse of the AMOC, the Atlantic's overturning circulation?
Main finding	In the simulations, collapse depended on the rate of warming, not a fixed temperature
In the model	Slow warming: still stable at ~5.5C. Warming at ~today's pace: collapsed near ~2C
Proposed mechanism (in the model)	Slow change lets the ocean adapt (saltier surface water, sea-ice changes); fast change overwhelms it
Method	One coupled climate model plus a simplified conceptual box model
Key caveat	Authors say the real AMOC is likely closer to tipping than their model; single idealized study

Frequently asked questions

What is the AMOC?

The Atlantic Meridional Overturning Circulation is a large system of ocean currents, including the Gulf Stream, that carries warm surface water northward and cold deep water southward through the Atlantic. It helps regulate temperatures, especially around the North Atlantic, so a major slowdown or collapse would have wide-ranging climate effects.

Does this study say the AMOC is safe up to 5.5C?

No, and the authors are explicit about that. The "still stable at 5.5C under slow warming" figure is a property of their model, which they say is too stable because of known biases. They state the real AMOC is probably closer to a tipping point than their simulation, so the result should not be read as a safety margin.

What is the actual new idea here?

That the rate of warming, not just the total amount, may decide whether the AMOC collapses. In the model, warming slowly to a given temperature could avoid collapse while warming quickly to the same temperature triggers it, because slow change gives the ocean time to adapt.

How confident should we be in this?

Cautiously. It is a single modelling study using one climate model plus a simplified conceptual model and idealized CO2 scenarios, not observational evidence or a multi-model consensus. Its value depends on whether other research groups reproduce the rate-dependent behaviour.

Why would the rate of warming matter?

In the model, the authors propose that the stabilising processes that can keep the current going, such as increased evaporation making surface water saltier and changes in sea ice, operate slowly. Under gradual warming they have time to work; under rapid warming the ocean cannot keep pace and the destabilising feedbacks dominate. Whether that mechanism operates the same way in the real ocean is one of the things other modelling groups still need to test.