The number worth keeping is 0.8C, the remainder when the report's most optimistic peak of 1.8C around mid-century is subtracted from the 2.6C that governments' current policies deliver by 2100 [3][4][16], and since one of the report's co-authors, Joeri Rogelj of Imperial College London, prices delay at a tenth of a degree of extra peak warming for every five years of it [7], that 0.8C is roughly forty years of delay stated in temperature [17]. It is an identity, not a forecast, built directly from the report's own two scenarios.
Closer in, the margin is one tenth of a degree, 1.4C now against the 1.5C line [2][18], which happens to be exactly what a single five-year delay adds to the peak [21]. The measurement convention matters more than it looks, and this reading is mine rather than the report's: the 1.4C figure is a 20-year average, not a single year's reading [2], so the year in which one year's temperature clears 1.5 and the year in which the 20-year index does are different years, and a fund mandate or an insurance trigger written against "1.5C" without naming its averaging window is exposed to whichever measure its counterparty picks after the fact.
What changes for allocation is duration. The route sketched runs through a peak and then gradual cooling to below 1.5, ideally by the end of the century [12], which is about half a century spent above the line even in the ambitious case [20], with co-author Richard Betts of the University of Exeter putting the overshoot at many, many decades [8]. Half a century outlasts any fund life, so the honest framing here is procurement rather than investment: the asset is a contracted stream of removal purchases that has to survive the tenure of everyone who signs it.
The weak point in this thesis is worth naming directly. UNEP issues reports, not purchase orders, and Bill Hare of Climate Analytics says the agency describes the hole we have dug well and does a poor job of showing a way out [10]. Described demand with no obligated buyer behind it prices at zero. The second failure mode is worse for the thesis and better for emitters, which is that overshoot language becomes the permission structure for deferral, Betts's "we have to live with this warmer world" [8] reading rather differently in a finance ministry than in a laboratory, and then Rogelj's tenth of a degree per five years [7] arrives as the invoice while the spending goes to adaptation hardware instead of removal capacity.
What would show me wrong is a statutory removal obligation, with volumes and a strike price, in any large emitter inside the next couple of years, a test that resolves long before any of the report's temperatures do [3].
