The number is 0.35°C per decade. Stefan Rahmstorf and Grant Foster at the Potsdam Institute published it in Geophysical Research Letters on March 6, and it should change how every engineer on the planet prioritizes their work. The warming rate since roughly 2015 is nearly double what it was from 1970 to 2015. Five independent datasets confirm it. After stripping out El Niño, solar cycles, and volcanic noise, the signal holds at greater than 98% statistical confidence. I think this is the clearest engineering deadline humanity has ever received, and we should treat it like one.
Some people want to argue about whether this proves we crossed a tipping point. I understand the instinct. Claudie Beaulieu at UC Santa Cruz is right that the acceleration could prove temporary. The Rahmstorf and Foster study itself does not identify a cause; reduced aerosol emissions from shipping regulations are a candidate, but unproven. Fair enough. The mechanism matters for climate science. It does not matter for what we do next.
The Signal Is the Schedule
Think about this the way a flight director thinks about a countdown clock. You do not need to know why the clock is running faster to know your timeline just got shorter. If the 0.35°C rate holds, the long-term 1.5°C Paris threshold gets breached somewhere between 2026 and 2029, depending on which dataset you trust. That is not a century away. That is before most car loans issued today get paid off.
The last 12 years are the warmest in the instrumental record. That includes La Niña years. It includes 2025, which started under cooling Pacific conditions and is still running hot. Munich Re's April 10 analysis forecasts a strong El Niño developing in the second half of this year, potentially pushing 2027 to a new record. The baseline keeps rising, and the spikes keep getting taller.
None of this means the game is over. It means the game has a clock now.
Build Faster
Electric vehicles hit 22% of global car sales in 2024, up from under 5% in 2020. Renewables reached 47.3% of electricity generation in Europe last year. China hit 33.6%. The US reached 24.1%, more than double its 2010 share. These are real deployment curves, driven by real engineers solving real manufacturing problems. They are also not fast enough if the warming rate stays at 0.35°C per decade.
The response to a shortened timeline is not despair. It is acceleration. Carbon removal needs to move from pilot scale to industrial scale on a timeline measured in years, not decades. Grid-scale storage needs the same iteration speed that got solar panels from $76 per watt in 1977 to $0.20 today. Permitting reform for transmission lines, geothermal wells, and nuclear plants needs to stop being a talking point and start being law.
I will grant the skeptics something: declaring a tipping point crossed before we understand the mechanism carries real risk. It can feed fatalism. It can make people stop trying. But the opposite failure is worse. Treating a 98%-confidence acceleration as background noise, waiting for perfect causal attribution before changing the build schedule, is the engineering equivalent of watching your fuel gauge drop and deciding you need a better theory of combustion before pulling into a gas station.
The teams building the next generation of electrolyzers, the crews pulling cable for offshore wind farms, the researchers testing direct air capture sorbents: they are working against this clock whether policymakers acknowledge it or not. The Rahmstorf and Foster paper did not create the deadline. It measured it.
We have built things fast before. The US went from zero nuclear reactors to over 100 in 25 years. China installed more solar capacity in 2023 than the entire world had in 2016. Speed is a choice. The 0.35°C number just made the consequences of choosing slow a lot more concrete.
