Short answer no, doesn’t seem possible on low emissions. Okay this is a preview of what the next IPCC report AR7 might say about AMOC collapse based on a Carbon Brief attempt at a survey of the full field of papers.
- None of them have collapse on our very low emissions path.
So that seems pretty much ruled out
- On low emissions AMOC collapse might be possible next century and take a century to fully collapse.
- On medium emissions it might start even this century in the 2060s, for that we probably have to exceed 2.5 C above preindustrial.
- However another study couldn’t get a collapse this century even on high emissions.
- All the models show it weakening, but collapse is much harder to trigger.
That’s based on my summary of one of their sections:
My summary of: What does the latest science say about when AMOC could ‘collapse’?
They found a range of predictions. One found limited weakening even on a very high (read impossible) emissions scenario.
Meanwhile, research published in Nature Geoscience in 2025 found that AMOC would experience “limited weakening” of 18-43% – or 3-6Sv – by 2100, even in a very-high emissions scenario. As such, it said an “AMOC collapse during the 21st century was unlikely”.
To reach this conclusion, the scientists refined existing climate model projections by developing a model based on ocean physics which took into account the relationship between water density and the AMOC’s depth. They also incorporated real-world measurements, including from the RAPID array.
https://interactive.carbonbrief.org/amoc-explainer/index.html#what-does-the-latest-science-say-about-when-amoc-could-collapse
Carbon Brief looked at a range of studies similarly to the way AR7 might do. This is the one that is closest to a collapse this century but it was on medium not low emissions:
QUOTE STARTS
Another recent study, published in JGR Oceans in 2025, found that the AMOC could “begin to collapse” as soon as 2063 under a medium-emissions scenario.
…
After analysing when this trigger point occurred in various model runs, they pinned the “AMOC tipping threshold” at around 2.5C of global warming above the pre-industrial average.
https://interactive.carbonbrief.org/amoc-explainer/index.html#what-does-the-latest-science-say-about-when-amoc-could-collapse
We are headed for 1.5 C with the triple renewables pledge for context and well below 2 C with just the net zero pledges for COP 26.
BLOG: We are ahead of target for 1.5 C with the COP28 triple renewables pledge
— media and academics lag far behind what’s happening in the real world when they claim we are headed for 2.6 C
You can read it here:
https://robertinventor.substack.com/p/world-is-overachieving-for-15-c-with
Then as for how long it takes then they say it could take 100 years.
QUOTE STARTS
“[Our research shows] that many [simulations to 2100] have already reached a critical value where AMOC has started to tip – a process that could then take 100 years. In those, the [simulated] AMOC might be at 12Sv by 2100, but actually it is already collapsing.”
https://interactive.carbonbrief.org/amoc-explainer/index.html#what-does-the-latest-science-say-about-when-amoc-could-collapse
So - short summary of these tentative results, it may take a century if it is possible and there’s a range of papers with conflicting results about whether it is possible on medium to high emissions next century, and one paper could get it to collapse slowly later this century at around 2.5 C.
Previous summary from AR6 - medium confidence that it’s certainly not going to collapse
Earlier in the article they summarize the results from AR6, it’s similar to what I explained before as I was summarizing the same results.
So here is my summary of AR6 which goes into a bit more detail than they do:
First, the AMOC isn’t the same as the wind blown Gulf stream which continues whatever happens. It’s below the surface and it MIGHT be slowing down but hard to be sure. There’s nothing to suggest a collapse, and what evidence we have is against the idea.
The AMOC can’t stop suddenly, it would take decades. Also as far as we can tell from the best models we have, it is impossible for it to stop this century. But it’s hard to model.
When the IPCC says
“There is medium confidence that there will not be an abrupt collapse before 2100. ”
- this is rather unintuitive to most readers.
Most people would read that as saying that it is quite likely to happen.
But no it means nothing like that.
Let’s take an example from racing. Suppose you want to know if the horse called Lucky Star will win the race. You have a couple of data points that it came last in two races so far this season. But you can’t find any more example of races it ran in and it might just have had two bad days in those races.
So then you’d say that it won’t win the race, with medium confidence because you have very little data but the data you have suggests it has no chance of winning. So medium confidence that Lucky Star has no chance of winning.
Medium confidence that it won’t win does NOT give any likelihood that it will win.
Indeed, if anyone wants to bet on Lucky Star, your best prediction is that it will lose. After all it was last twice in succession and you know nothing about it to suggest it can beat the other horses.
So you have medium confidence that it is going to lose.
TEXT ON GRAPHIC: Why Medium confidence that the AMOC will NOT collapse does not give any likelihood that it WILL collapse.
Placing last twice is NOT a good reason to place ANY bet on Lucky star to win
Medium confidence that Lucky Star will NOT win based on 2 previous races where it came last both ties does NOT translate into ANY level of likelihood that Lucky Star wins
Graphic made with Bing Copilot
Similarly with the AMOC, all the realistic models the IPCC looked at for AR6 said a collapse is impossible at our current warming level.
But they found it hard to model the ice melt from Greenland. It is difficult to get it to stop even if they simulate melting more ice than there was during the last ice age. So there is something they are missing because we know it did stop several times coming out of the last ice age, with far more fresh water than can be melted today.
They expect that once they can model it properly the models will still say it is impossible with the small amount of potential melt water from Greenland. But they can’t currently do this modeling.
So it is medium confidence that it is impossible.
They don’t add in a likelihood level but if you added it in, since it never stops in the models you’d put it like this:
“There is medium confidence that [it is certain that] there will not be an abrupt collapse before 2100. ”
It reads a bit odd but that would be a fair summary of the research on the AMOC as summarized in AR6. There is NO confidence that it WILL collapse. Because they can’t model that.
Modelers have been able to get the AMOC to stop but at least as of AR6 none of their successful models of it stopping were realistic.
We know it CAN stop in the real world as it did at times as we came out of the last ice ages but that was with far more fresh water melt than is possible today. We can’t model those times either though which is why the models are of somewhat limited use.
[I don’t know if any of these newer models can now model the times that we know it did stop]
Generally though if it was going to stop you’d expect it is more likely with more warming. On our current low emissions path we only warm by another fifth of a degree or so and there is nothing to suggest it is even headed for a partial collapse at present.
BLOG: We are ahead of target for 1.5 C with the COP28 triple renewables pledge
— media and academics lag far behind what’s happening in the real world when they claim we are headed for 2.6 C
You can read it here:
https://robertinventor.substack.com/p/world-is-overachieving-for-15-c-with
So - if Lucky Star winning is the AMOC stopping, it would make no sense to bet on Lucky Star.
Here is where the article you shared explains AR6 but not very clearly:
In AR6, the IPCC notes it has “medium confidence” that the decline of AMOC will not involve an “abrupt collapse” before 2100. (An “abrupt” change in IPCC lingo is an event taking place in three decades or less.)
The IPCC’s findings were backed up by a 2025 Nature study that examined the future stability of the AMOC in 34 climate models adjusted to simulate varying levels of greenhouse gas emissions and freshwater input.
The researchers found that an AMOC collapse – defined in a correction notice as a “weakening to below 6Sv” – was “unlikely” this century, noting that, “in all cases”, the ocean circulation was sustained “by upwelling in the Southern Ocean, driven by persistent Southern Ocean winds”.
https://interactive.carbonbrief.org/amoc-explainer/index.html#what-does-the-latest-science-say-about-when-amoc-could-collapse
BLOG: Nothing to suggest AMOC will stop though likely is slowing down
— would take decades to stop
— Gulf Stream continues anyway
— much misunderstood term “medium confidence” in IPCC reports
You can read it here:
https://doomsdaydebunked.miraheze.org/wiki/Nothing_to_suggest_AMOC_will_stop_though_likely_is_slowing_down_-_would_take_decades_to_stop_-_Gulf_Stream_continues_anyway_-_much_misunderstood_term_%22medium_confidence%22_in_IPCC_reports
So this article says that some of the new models can now get it to collapse this century or next. So that is progress.
What I’d like to know is how realistic those models are thought to be and especially how well these models do with modelling the times in the past when we know the AMOC did collapse as the ice sheets melted coming out of the last ice ages.
If they can get it to collapse in the past, and the details also resemble what really happened in the past to some degree, it would suggest we are on the right lines and can be more accurate in our results.
Carbon Brief doesn’t go into that, nor do any of the papers by their titles. But AR7 surely will do as it will be far more thorough. And Carbon Brief doesn’t give any confidence level for the new results. Are they also medium confidence or are we getting to a higher level of confidence now? AR7 will do that too.
We’ll get AR7 some time around 2029. Discussed it here:
https://ddebunked.org/d/4152-a-bit-of-a-weird-one/6
Anyway all of this suggests that on low emissions it’s way below that possible 2.5 C tipping point where it might eventually collapse over a period of a century in their most sensitive model runs.
So that sounds like good news on AMOC collapse.
As for the section about effects, remember it’s about effects that unfold slowly over a century probably next century and AR7 will likely show those effects are impossible on low emissions. And if the AMOC does influence some of those other effects it just starts a process that in many of the examples of things it could possibly tip over would then would unfold over additional centuries to millennia too.
And by then we have far future technology to handle whatever happens. The issue with the climate change is mostly rapid climate change. We, and our ecosystems can adapt to slow changes that take centuries. Nature and our civiization could easily have adapted to the 1.2 C rise since pre-industrial if it was 0.2 C per century instead of the most recent 0.2 C per decade. It’s been rising at an average of just under 0.1 C per century since preindustrial. At a tenth of the rate it would have taken well over 1000 years to get here and we’d have adapted easily.
BLOG: The warming of the Anthropocene has benefited the world in many ways
— the issue is the speed of change and we likely wouldn’t want to go rapidly back to preindustrial from 1.8 C
You can read it here:
https://debunkingdoomsday.quora.com/The-warming-of-the-anthropocene-has-benefited-the-world-in-many-ways-the-issue-is-the-speed-of-change-and-we-likely-wo
