The Atlantic Ocean’s most powerful circulation system may be closer to a dangerous tipping point than mainstream climate models have suggested, according to a body of peer-reviewed research published between 2023 and early 2026 that has sharpened a long-running scientific debate. The latest findings, including a February 2026 paper identifying abrupt shifts in the Gulf Stream’s path as a potential early warning signal, have pushed the discussion from theoretical risk toward active monitoring, with consequences that could reshape weather, agriculture, and sea levels for billions of people across three continents.
The system in question is the Atlantic Meridional Overturning Circulation, or AMOC, a vast conveyor belt of ocean currents that carries warm tropical water northward along the surface and returns cold, dense water southward at depth. It is one of the planet’s primary heat-redistribution engines. Without it, Northern Europe would lose a significant source of warmth, the West African monsoon could shift, and sea levels along the U.S. East Coast could rise by as much as half a meter on top of global averages, according to projections cited in multiple studies.
What the latest research shows
The sharpest warning came from a 2024 paper in Science Advances led by Rene van Westen at Utrecht University. Using a high-resolution climate model, the team identified a metric based on freshwater transport at the southern boundary of the Atlantic basin. When enough fresh water from Greenland’s melting ice sheet and increased rainfall dilutes the salty North Atlantic, seawater becomes too light to sink, and the deep-water formation that drives the AMOC stalls. Van Westen’s team validated their model against observational reanalysis data and concluded the circulation is already moving toward the regime where that feedback loop could become self-reinforcing.
“Once you pass the tipping point, the AMOC collapse is irreversible on human timescales,” van Westen told reporters when the paper was published, a statement that drew attention precisely because the metric his team proposed is physically grounded rather than purely statistical.
That work built on a 2023 study in Nature Communications by Peter and Susanne Ditlevsen, who applied statistical early warning methods to sea-surface temperature proxies sometimes called an “AMOC fingerprint.” Their analysis estimated a potential tipping window with a central estimate around 2057 and a 95 percent confidence interval stretching from 2025 to 2095. The lower bound of that window has now arrived, which is one reason the scientific conversation has shifted from abstract future risk to near-term vigilance. The Ditlevsen study’s statistical indicators showed that variability patterns in the North Atlantic are changing in ways consistent with a system losing resilience, though the authors cautioned that their proxy-based approach carries inherent uncertainty.
A February 2026 paper in Communications Earth and Environment added a new dimension. Researchers identified abrupt shifts in the Gulf Stream’s path as a dynamical precursor to AMOC collapse. Unlike older proxies that relied on broad temperature patterns, this study tied a directly observable feature of the Gulf Stream to modeled stability loss in the overturning circulation. This paper was published recently and its findings have not yet been widely replicated. If confirmed by ongoing satellite and buoy observations, Gulf Stream path changes could serve as an independent early warning signal alongside freshwater transport data, giving scientists two separate physical indicators to track.
The observational backbone for much of this work is the RAPID-MOCHA-WBTS array, a trans-basin monitoring system deployed at 26.5 degrees north latitude since 2004. Operated jointly by the U.K.’s National Oceanography Centre and the U.S. National Oceanic and Atmospheric Administration, the array is widely considered the gold standard for measuring the AMOC’s strength and structure, according to NOAA’s Atlantic Oceanographic and Meteorological Laboratory. Its continuous records show a weakening trend over the past two decades, though that signal sits on top of substantial year-to-year variability that complicates interpretation.
Why scientists still disagree
Not all research points to imminent danger. A 2024 study in Nature led by Katinka Bellomo examined 34 climate models run under SSP5-8.5, the highest-emission scenario used in international assessments. The models projected significant AMOC weakening but did not produce an abrupt shutdown before 2100. That finding stands in direct tension with the observational and proxy-based early warning studies, and the gap between the two camps has not been resolved.
One reason for the disagreement is structural. Many current-generation climate models may not adequately capture the small-scale mixing processes and freshwater feedbacks that drive tipping behavior. The freshwater transport mechanism highlighted by van Westen’s team operates through a feedback loop that some models handle incompletely, either because of limited ocean resolution or simplified representations of ice-sheet melt and Arctic freshwater export. If those models underestimate the speed at which the North Atlantic freshens, they could also underestimate tipping risk.
Proxy-based approaches carry their own limitations. The sea-surface temperature fingerprint used by the Ditlevsen team infers AMOC behavior indirectly. Temperature patterns at the ocean surface can be influenced by wind changes, atmospheric variability, and other currents, which makes isolating the AMOC signal difficult. Statistical early warning indicators like rising autocorrelation can flag a system that is slowing down, but they do not guarantee a true tipping point is near.
A separate line of research by Sybren Drijfhout, published in Environmental Research Letters and summarized in a press release from the journal, used extended runs of IPCC-standard models to argue that a North Atlantic overturning shutdown could occur after 2100 under high-emission futures. Drijfhout’s work challenges the IPCC’s earlier characterization of AMOC collapse as a “low-likelihood” event, showing that a meaningful fraction of model runs eventually produced a shutdown when simulations were extended beyond the usual 2100 cutoff. The probability of collapse, the researchers argued, is no longer negligible, though the precise thresholds depend heavily on which emission pathway the world follows.
The RAPID-MOCHA array, while invaluable, covers only about two decades of continuous data. That record is too short to distinguish confidently between natural multi-decadal variability and a long-term trend toward collapse. Researchers have not yet integrated the newer Gulf Stream path precursor data from the February 2026 study into the array’s monitoring framework, and as of May 2026, no public statements from NOAA address how or whether that integration will proceed. The observational network is still evolving, and conclusions drawn today may be revised as longer records accumulate.
What collapse would actually mean
The stakes are not abstract. If the AMOC were to weaken sharply or shut down, Northern Europe would lose a heat source that keeps winter temperatures several degrees Celsius warmer than they would otherwise be at those latitudes. Agricultural growing seasons in the U.K., Scandinavia, and parts of France could shorten. The West African monsoon, which hundreds of millions of people depend on for rain-fed farming, could shift southward. Sea levels along the U.S. East Coast, already rising faster than the global average in part because of AMOC-related dynamics, could accelerate further, threatening cities from Miami to Boston.
These consequences are why the debate over timing matters so much. A collapse in the 2040s would demand emergency-scale adaptation. A collapse after 2200 would still be catastrophic but would allow more time to prepare. The current evidence does not resolve which scenario is more likely, and that ambiguity is itself a policy problem.
How competing evidence shapes the monitoring challenge ahead
Readers trying to make sense of competing headlines should distinguish between three types of evidence in this debate. Direct observations from instruments like the RAPID-MOCHA array provide the most trustworthy data on what the AMOC is doing right now. Those measurements show a weakened circulation compared with the early 2000s, but they do not prove a point of no return has been crossed.
Physics-based early warning signals, including the freshwater transport metric and the Gulf Stream path precursor, offer mechanistic reasons to expect trouble. They highlight feedbacks that could accelerate weakening once certain thresholds are passed, turning a gradual decline into an abrupt transition. But these indicators depend on models to connect present-day measurements to future tipping points, and small changes in assumptions can shift the inferred timeline by decades.
Large model ensembles, like the 34-model study in Nature, test whether simulated oceans collapse under extreme warming. Their results are only as reliable as the physics coded into each model, but they represent the broadest available survey of how independent modeling groups expect the system to behave. When most models agree that collapse before 2100 is unlikely, that consensus carries weight, even if it may underestimate risks tied to processes the models handle poorly.
The practical tension is clear: the models best equipped to simulate long-term climate generally do not show abrupt AMOC collapse this century, while observational and proxy-based studies suggest the system is already drifting toward a tipping threshold. For policymakers, that creates a difficult risk-management problem. Waiting for certainty could mean discovering too late that the AMOC was closer to failure than models implied. Acting on the most alarming scenarios means investing heavily in adaptation for an event that may not arrive for generations.
Most researchers frame AMOC collapse as a low-probability but high-impact risk in the near term and a growing concern on multi-century horizons. Cutting greenhouse gas emissions reduces the chance of crossing critical thresholds. Expanding ocean monitoring networks improves the odds of catching early warning signs before they become irreversible facts. As of spring 2026, the science has not delivered a verdict, but it has delivered an ultimatum: the window to keep the worst outcomes unlikely is narrowing, and the Atlantic will not wait for the models to catch up.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
