The Atlantic's Vital Current Is Weakening. Scientists Are Split on What Happens Next.

In early June 2026, a team led by Stefan Rahmstorf at the Potsdam Institute for Climate Impact Research published findings in Geophysical Research Letters confirming what oceanographers had suspected for years: the persistent patch of cold water in the North Atlantic — the so-called "cold blob" — is not an atmospheric anomaly. It is a deep ocean phenomenon, driven by reduced heat transport from a weakening Atlantic Meridional Overturning Circulation (AMOC) [1]. The study, drawing on reanalysis data from 1955 to 2022 and sea-surface temperature records stretching back to 1880, ruled out surface heat loss as an explanation and pointed directly at declining circulation strength [2].

The AMOC is one of the largest current systems on Earth. It carries warm surface water northward from the tropics, where it cools, becomes denser, sinks to the deep ocean, and flows back southward — a conveyor belt that moves roughly 18 million cubic meters of water per second (measured in Sverdrups, abbreviated Sv) and distributes enormous quantities of heat to the Northern Hemisphere [3]. Its weakening has implications for sea levels along the U.S. East Coast, rainfall across Africa and South Asia, European agriculture, and hurricane behavior. The question dividing the scientific community is not whether the AMOC is slowing — it is — but whether it is approaching a threshold beyond which the slowdown becomes self-reinforcing and, on human timescales, irreversible.

Measuring the Slowdown

The primary observational tool for tracking AMOC strength is the RAPID 26°N mooring array, a line of instruments deployed between Florida and the Canary Islands since 2004. The array measures the volume of water transported northward at latitude 26°N [4].

Source: RAPID/Met Office

Data as of Dec 1, 2025CSV

The data shows a decline from roughly 18.7 Sv in 2004 to approximately 15.5 Sv in 2023, a rate of about 1 Sv per decade [4][5]. A sharp dip in 2009-2010, when strength fell to around 13.5 Sv, was later attributed primarily to wind forcing rather than long-term climate trends [4]. But the broader downward trajectory has persisted.

Proxy records extend the picture further back. Multiple studies using ocean sediment cores, coral records, and ice-core data indicate that the AMOC is now weaker than at any time in at least 1,000 to 1,600 years [6][7]. A 2026 study published in Science Advances, applying observational constraints to climate model projections, calculated that the AMOC could lose 51% of its current strength by 2100 under a medium-emissions pathway — substantially more than the 24-39% range projected by the IPCC's Sixth Assessment Report [8][9].

The Cold Blob as Fingerprint

The North Atlantic cold blob — an area south of Iceland and Greenland where sea-surface temperatures have dropped against the global warming trend — has become the most visible signature of AMOC weakening. Rahmstorf's June 2026 paper demonstrated that the cooling is not caused by increased heat loss from the ocean surface to the atmosphere, which has actually declined since 1993. Instead, the ocean is receiving less heat from the south because the circulation carrying that heat has weakened [1][2].

High-resolution fingerprint analysis published in October 2025 on RealClimate further reinforced this interpretation, showing that the spatial pattern of North Atlantic cooling matches what models predict from reduced overturning circulation, not from other potential causes [10].

The Freshwater Driver

The leading mechanism behind AMOC weakening is freshwater input from the melting Greenland ice sheet and increased Arctic river discharge. Fresh water is less dense than salt water. When it enters the North Atlantic in sufficient quantities, it disrupts the sinking of cold, dense surface water — the engine that drives the overturning circulation [11].

Source: NSIDC / GRACE

Data as of Jun 1, 2025CSV

The scale of Greenland's ice loss is substantial: approximately 5,900 gigatons since 2002, as measured by the GRACE satellite mission [11][12]. Between 1985 and 2020, the ice sheet lost roughly 4,200 gigatons [11]. This meltwater, entering the North Atlantic, has measurably reduced surface salinity in the regions where deep-water formation occurs.

A key question is the threshold at which freshwater input crosses from causing gradual weakening to triggering collapse. Hosing experiments — in which climate models are subjected to artificial freshwater inputs — suggest that some models lose AMOC stability at inputs as low as 0.3 Sv, while others remain stable at higher levels [13]. This model-to-model inconsistency is one of the central sources of scientific disagreement about collapse risk.

The Tipping Point Debate

The phrase "tipping point" implies a specific physical property: bistability. In a bistable system, the AMOC has two stable states — "on" (current strength) and "off" (collapsed) — separated by a threshold. Once that threshold is crossed, the system shifts to the collapsed state and cannot return to full strength without a reversal of the conditions that caused the transition [14].

Whether the real-world AMOC has this bistability is contested. In most CMIP6 models — the current generation of full-complexity climate simulations used by the IPCC — the AMOC sits in a monostable regime, far from any tipping point. Researchers attribute this to "subtle biases in the Atlantic salinity distribution" within the models, suggesting they may be systematically too stable [13][15].

On the other side, intermediate-complexity models and some observational analyses support the bistability hypothesis. A March 2025 study reported "reconciled warning signals in observations and models" that imply an approaching tipping point [16]. Susanne Ditlevsen and Peter Ditlevsen, in a widely cited 2023 Nature Communications study, used statistical analysis to estimate a most likely collapse date around 2065, with a 95% confidence interval of 2037 to 2109. An August 2025 update shifted their mean estimate to 2050, with a 59% probability of collapse before that date [17].

Critics of these projections argue that the intermediate-complexity models used are "considered less reliable in general and may confuse a major slowing of the circulation with its complete collapse," as the Science Media Centre reported in its expert roundup [18]. A February 2025 study published in Nature concluded that the AMOC is "resilient" to extreme greenhouse gas and freshwater forcings across 34 climate models, pushing back against collapse narratives [19].

Laura Jackson of the UK Met Office has noted that public perceptions of AMOC collapse are shaped by dramatized scenarios. "What we are thinking about happens over decades or maybe a century," she said, not the sudden shutdown depicted in popular media [4].

The IPCC Gap

The divergence between the IPCC's AR6 assessment and more recent studies is one of the most consequential scientific disputes in climate research.

AR6, published in 2021, stated with "medium confidence" that abrupt AMOC collapse before 2100 is unlikely, projecting a 24% decline under low-emissions scenarios and 39% under high emissions [9]. The report assigned "low confidence" even to the question of whether the AMOC had weakened during the 20th century — partly because CMIP5 models showed a slowdown while CMIP6 models showed a strengthening, an inconsistency that led the IPCC to revise its confidence downward [13].

Since AR6, the balance of evidence has shifted. Researchers at Stellenbosch University have argued that the probability of AMOC collapse before 2100 is "very likely to be underestimated in IPCC-AR6 and needs to be reconsidered in the IPCC-AR7" [8]. A 2025 paper in the Journal of Geophysical Research: Oceans suggested tipping could begin at approximately 2.5°C of global warming, with collapse onset around 2063 [20]. A separate 2025 study in Environmental Research Letters found that 67% of model runs showed post-2100 shutdown under very high emissions [21].

The methodological gap comes down to model complexity and observational constraints. The IPCC relies heavily on CMIP6 models, which tend to show a stable AMOC. Studies projecting earlier collapse either use simpler models with more explicit bistability or apply observational filters that select for models closer to real-world conditions — which tend to show more vulnerability.

Source: OpenAlex

Data as of Jan 1, 2026CSV

The surge in AMOC-related publications — from 197 papers in 2011 to 1,306 in 2025, with over 9,200 total — reflects the intensity of this scientific effort.

Who Gets Hurt: Coastal Flooding, Drought, and Crop Failure

The consequences of significant AMOC weakening — even short of full collapse — are geographically uneven.

U.S. East Coast sea level rise. AMOC collapse would redistribute ocean mass, adding an estimated 50 centimeters or more of sea level rise along the eastern United States on top of warming-driven increases. Jupiter Intelligence, a climate risk analytics firm, analyzed over 16,000 properties valued at $14.1 billion and found that AMOC collapse would increase 100-year flood exposure by a factor of 2.8. Miami and South Florida face projected losses of nearly $500 million; the New York City metro area, $270 million. More than 100 ZIP codes across 12 U.S. states that currently face minimal coastal flood risk would become newly exposed [22]. For context, Hurricane Sandy's damage to the New York area included approximately $8 billion in losses attributed to just 10 centimeters of climate-driven sea level rise [22].

African Sahel drought. A weakened AMOC shifts the Intertropical Convergence Zone (ITCZ) — the planet's primary tropical rain belt — southward, because the AMOC's northward heat transport helps position the ITCZ farther north. A 2024 study by Ben-Yami et al. in Earth's Future found remarkable agreement across four independent climate models: the West African Monsoon would see approximately 29% annual rainfall decline, the Indian Summer Monsoon 19%, and the East Asian Summer Monsoon 4%. These impacts persist for at least 100 years [23]. The Sahel, home to hundreds of millions of people dependent on rain-fed agriculture, could transition from semi-arid climate to arid desert [24].

European temperature shifts. Reduced Atlantic heat delivery would cool northwestern Europe, though the relationship is not straightforward. Some research suggests that while AMOC weakening cools the subpolar North Atlantic, it could paradoxically favor heatwaves in parts of Europe by altering atmospheric circulation patterns [25].

The Uncomfortable Question: Who Benefits?

Media coverage of AMOC weakening is almost uniformly framed as a threat. But some research points to regional countereffects that rarely surface in public discussion.

A weakened AMOC cools North Atlantic sea-surface temperatures. Since warm ocean water is a primary fuel for hurricanes, some researchers have hypothesized that reduced AMOC heat transport could dampen hurricane intensification in parts of the Atlantic. However, paleoclimate evidence complicates this narrative: sediment records from the Younger Dryas period (roughly 12,900 to 11,700 years ago), when the AMOC was substantially weaker, show that environmental conditions near Florida actually favored stronger storm development despite cooler sea-surface temperatures [26][27].

In the Southern Hemisphere, AMOC collapse produces a striking paradoxical effect. The same models that project severe Sahel drying show the Southern Amazon receiving over 43% more annual rainfall [23]. Whether this constitutes a "benefit" depends on the magnitude and seasonal distribution — excessive rainfall can cause flooding and ecosystem disruption.

The question of whether the absence of these nuances in mainstream reporting reflects advocacy or editorial prioritization is itself a subject of debate. Scientists studying AMOC emphasize that the net global effects of weakening are overwhelmingly negative, and that highlighting isolated regional countereffects risks obscuring the systemic nature of the risk.

Paleoclimate Precedent: Dansgaard-Oeschger Events

The AMOC has collapsed before — many times. During the last glacial period, repeated episodes known as Dansgaard-Oeschger (D-O) events saw abrupt Northern Hemisphere warmings of 8 to 15°C within decades, driven by AMOC resurgences from a weakened or collapsed state. These events followed an approximate 1,470-year periodicity and demonstrated the "bipolar seesaw": when the AMOC strengthened, Greenland warmed rapidly while Antarctica cooled, and vice versa [28][29].

The relevance to today is debated. D-O events occurred under glacial conditions fundamentally different from the present interglacial — with massive Northern Hemisphere ice sheets, lower sea levels, and different atmospheric CO₂ concentrations. The Late Pleistocene AMOC was far more volatile than the Holocene AMOC has been [28]. Some researchers argue this means the current system is more stable; others counter that the unprecedented pace and nature of modern freshwater forcing (from anthropogenic warming) is pushing the system into territory with no Holocene analogue [30].

The key distinction: D-O events show that the AMOC can indeed cross tipping points and shift abruptly between states. Ecosystems and human civilizations did eventually adjust — but over centuries to millennia, and under conditions where global population was a tiny fraction of today's 8 billion.

What Scientists Agree On — and Where They Don't

Points of broad consensus: the AMOC has weakened; Greenland meltwater is a significant driver; continued weakening under further warming is very likely; the consequences of major weakening would be severe for hundreds of millions of people [6][9][23].

Points of sharp disagreement: whether the AMOC has true bistability in the current climate; whether intermediate-complexity models or CMIP6 models better capture the risk; whether collapse is a this-century concern or a next-century one; and whether the RAPID array's 20-year record is sufficient to distinguish anthropogenic signal from natural variability [13][14][15].

Niklas Boers of the Potsdam Institute has framed the central question: "The AMOC is tipping — in the sense of, can it come back or not? — remains the practically relevant question" [15]. Stefan Rahmstorf has acknowledged a fundamental constraint on certainty: "We are never going to get a reliable 'early warning'" because the observational infrastructure and data record are not yet mature enough [10].

The RAPID array, scientists estimate, will need roughly 29 years of continuous data — until approximately 2033 — to reliably isolate human-caused signals from natural variability [4]. Until then, the debate will continue to be shaped as much by model assumptions and statistical methods as by direct observation.

What is not in dispute is the scale of what is at stake. The AMOC redistributes heat that sustains agriculture, moderates weather, and stabilizes sea levels across two hemispheres. Its weakening, however gradual or abrupt, represents a structural shift in the climate system — one that is already underway.