この責任は、部門の規模に見合っています。UNEPとGlobal Alliance for Buildings and Construction(GlobalABC)が2026年5月に公表した『Global Status Report for Buildings and Construction 2025-2026』が、その位置を示しています。UNEPのInger Andersen事務局長は、建物は気候リスクを固定することもできるし、より安全で健康的で手頃な暮らしを実現することもできると述べています。同報告書は、2050年に存在する建物のおよそ半分が、これから新築されるか、2050年までに改修されるものになると見込んでいます。
Sooner for buildings, wider for society, heavier for those who invest in real estate
In 2015 the Paris Agreement set the goal of holding the rise in global average temperature well below 2°C above pre-industrial levels, and of keeping it to 1.5°C so far as possible. In the decade since, national reduction targets and corporate decarbonisation plans alike have been built around that 1.5°C figure.
Exceeding 1.5°C is now, however, no longer avoidable. Limiting Overshoot, published by the United Nations Environment Programme (UNEP) on 2 September 2026, is written on that assumption. UNEP's position is that exceedance is not an argument for resignation but a reason to press harder. Greenhouse gases that were not cut remain in the atmosphere, and the effects of the resulting warming arrive sooner and reach further than expected. On the reading given by the International Court of Justice in 2025, the legal duty on states to regulate emissions also becomes more demanding as a result of exceedance, not less.
Limiting Overshoot is a 139-page document addressing not how to avoid crossing 1.5°C but what to do on the assumption that it is crossed. This article reads that report from the position of those who own and manage buildings. Except where stated otherwise, what follows quotes or summarises that report. A glossary of the main terms appears at the end.
September 2026 Yukihiko Ito, Founder and CEO, GOYOH Inc.
01
Where we now stand
The report opens with a set of indicators. In 2024, annual global mean temperature exceeded 1.5°C for the first time in a single calendar year, and the world has since entered a 20-year period whose mean warming is 1.5°C. Mass coral bleaching was documented in at least 83 countries and territories, and in February 2025 combined Arctic and Antarctic sea ice reached a new all-time minimum. Global glacier loss has concentrated in the decade from 2015 to 2024, with 6 per cent lost in 2023 alone.
Each square represents 2 per cent of the whole. Coral: the share of tropical reefs that experienced bleaching between 2023 and 2025. Sea ice: the share of the oldest and thickest summer-surviving ice lost since the 1980s. Population: the share that experienced at least 30 days of extreme heat between May 2024 and May 2025. Source: UNEP, Limiting Overshoot (2026).
The human record is equally clear. Extreme heat here means days hotter than 90 per cent of temperatures observed locally over 1991 to 2020. Heat-related deaths during a European summer heatwave in 2025 were estimated to have tripled, with that increase attributed to global warming.
The remaining budget is starker still. To have a 50 per cent chance of limiting warming to 1.5°C, only about 130 billion tonnes of carbon dioxide can be emitted from 2026 onwards, which is about three years of emissions at the 2024 rate. Staying within it would require global carbon dioxide emissions to decline linearly from 2025 levels to zero in just over six years. Under current policies, median projected warming reaches around 2.6°C by 2100, within a range of 1.9°C to 3.6°C.
Key figures on where we stand
about 1.4°Chuman-induced warming to date, above pre-industrial levels
about 0.25°C per decadethe observed rate at which global temperature is rising
about 0.05°C per decadethe rate at which warming may come down under net-negative emissions
about 50 yearsto reverse a decade of warming at the current rate, on optimistic assumptions about carbon removal
How long this state lasts is also set out in the report, which divides the path from crossing 1.5°C to returning below it into four stages: the exceedance stage, as warming rises past 1.5°C; the peak stage, when the rise levels off and holds at its highest level; the decline stage, as warming comes back down; and stabilisation. Those are the terms used from here on.
Warming stops rising at the point where global emissions reach net zero. The report states that halting warming at any level below 2°C requires global CO2 emissions to reach at least net zero, and that coming close to 1.5°C requires reaching that point in the 2050s. The peak stage is the period during which the rise has stopped as a result.
Nor does temperature come down of its own accord. It falls only where people remove more carbon dioxide from the atmosphere than is emitted, and only where that net-negative state is sustained for a long time. The decline stage is the period during which that work is under way.
These three stages — exceedance, peak and decline — are therefore not a forecast of what will happen. The shape appears only on the assumption that net zero is reached in the 2050s and that the world then goes further, into net negative. Without that, there is no peak: under current policies, warming in 2100 is projected at around 2.6°C in central estimates and is still rising.
That path is what the report calls a pathway: a storyline of how emissions, and the warming they cause, move over the coming decades. It treats this pathway as a sequence that runs in decades, and states that while the duration of overshoot can vary, it typically lasts for at least several decades. The pathway extends to the end of the century or beyond, and even under optimistic assumptions about temperature decline it would take multiple decades before global temperatures return below 1.5°C. Because the appropriate response changes from stage to stage, this distinction carries the second half of the argument. A schematic of the timeline appears in the section on asset management taking different action at each stage.
One sentence here carries particular weight for real estate. Adaptation, in this article, means avoiding and reducing the damage that climate impacts cause. According to the report, adaptation decisions in most sectors are taken on planning horizons shorter than these decades, so the overshoot period falls outside what those decisions consider. The report names long-lived infrastructure planning as the notable exception.
Buildings and their plant are in service for decades, which places them squarely within that long-lived infrastructure. Real estate falls inside that exception. The decades of overshoot sit within the life and refurbishment horizon of a building. The report adds that adaptation needs and costs will largely be determined by climate risks at peak warming, with further needs arising during long-term reversal and stabilisation.
02
What we have lost
Much of what has been lost does not return when the temperature does.
The report states that 1.5°C of warming is already dangerous. At that level, the IPCC's Sixth Assessment Report identified high to very high risks for unique and threatened systems including coral reefs and other biodiversity hotspots, and for the cryosphere including the Arctic and mountain glaciers. Very high risk, in that assessment, denotes severe impacts and the presence of significant irreversibility or the persistence of climate-related hazards, combined with limited ability to adapt owing to the nature of the hazard or impact.
Limits to adaptation are already being reached. The IPCC identified soft adaptation limits already being encountered, particularly for the most vulnerable communities. The documented widespread bleaching of tropical coral reefs indicates that in a 1.5°C world, some ecosystem-based adaptation measures will increasingly reach hard adaptation limits.
What has been lost, above all, is optionality. The report states that delayed action has pushed the lowest avoidable peak warming above 1.5°C. Options for deep emissions reductions from 2025 onwards remain available, but a decade of insufficient action and persistently high emissions has raised the minimum level of peak warming.
03
How this differs from what was originally expected
According to the report, the most stringent emission reduction pathways available around the time of the Paris Agreement are no longer feasible: they required immediate, deep and sustained reductions beginning as early as 2015 or 2020. And even under those conditions, most studies indicated that warming would temporarily exceed 1.5°C before returning below 1.5°C by 2100.
Overshoot, in other words, is not an accident that has just occurred. It was priced into the Paris-era pathways from the outset. What has been lost is not a future without exceedance, but a future in which exceedance was shallow and brief. The distinction matters in practice. Treating this as the failure of a target, or as a deterioration within the expected range, leads to different next moves.
What was expected for 2030, and what is projected. The 2030 greenhouse gas benchmark consistently implied by the IPCC and UNEP is 40 GtCO₂e per year or lower. Assuming full implementation of the third-round NDCs as submitted by September 2025, global emissions in 2030 are projected to decline only slightly from their record 2024 levels, to around 51 GtCO₂e per year. Source: UNEP, Limiting Overshoot (2026).
How far has the gap opened? More recent studies accounting for delayed action, institutional constraints and technological feasibility show that even stringent pathways starting around 2025 result in best-estimate peak warming exceeding 1.5°C by about 0.2°C. The more than 400 billion tonnes of carbon dioxide emitted since 2015 have already raised the lowest achievable warming by about 0.2°C. Each additional five years of sustained high emissions adds approximately 0.1°C to peak warming.
Peak warming by scenario, and warming in 2100 under current policies. Points mark central estimates and bands mark ranges. A stringent pathway beginning in 2025 is still estimated to peak above 1.5°C. UNEP's most optimistic scenario assumes implementation of all nationally determined contributions and long-term net-zero targets. Source: UNEP, Limiting Overshoot (2026).
A second divergence lies in the models used for investment decisions. The report notes that financial models assume warming that only rises — in project lifetimes, return calculations and risk pricing alike — and that a pathway which peaks and returns therefore needs purpose-built analysis.
On the mitigation side, meaning the work of limiting climate change itself by reducing greenhouse gas emissions, the instruments have matured. The leading one is CRREM, the Carbon Risk Real Estate Monitor, which began as an EU Horizon 2020 research project and is now run by the CRREM Foundation, an independent non-profit based in the Netherlands. It sets out, year by year, the operational emissions and energy use consistent with a given climate target. CRREM's decarbonization pathways, published free of charge, translate global carbon budgets derived from IPCC scenarios into year-by-year intensity targets — carbon and energy per square metre of floor area — for each building type in each covered country. Their output is the Misalignment Year: the first year in which a building's projected carbon or energy intensity exceeds the pathway. That is not a prediction about whether the world holds 1.5°C; it is timing information about when this asset departs from the pathway, and it remains useful for sequencing retrofits even once overshoot is assumed.
The CRREM Foundation stated in July 2026 that its pathway update will reintroduce a 2°C scenario. The updated pathways are to be released for consultation in June 2027, with go-live expected in November 2027.
Assessment during an overshoot period nonetheless contains a question distinct from alignment: whether this building and this locality can withstand the height and the duration of the peak. CRREM's own FAQ states that it addresses transition risk and points users to other tools for physical risk.
04
Which social capital will be most eroded
The report sets out the impacts of the overshoot period domain by domain. What follows is ordered by proximity to real estate practice.
Access to water. Impacts on water systems are described as complex and partially irreversible: hydrological systems show hysteresis, so when temperature comes back to a given level, water supplies do not come back to the state they held at that level. The report attributes this to changes in weather patterns, in water flows, and in how people use and manage water. More than 2 billion people already lack safe water and 3.5 billion lack safely managed sanitation, and a temporary overshoot is expected to make those gaps harder still to close. One observation carries particular weight for practice: water infrastructure can last for decades, so systems built during a temporary warming period continue to affect communities and ecosystems long afterwards.
Access to energy and its affordability. The report treats cooling demand as a risk in its own right. Global electricity use for cooling could rise from nearly 5,000 terawatt-hours in 2022 to as much as 18,000 terawatt-hours by 2050 on a high-emissions pathway. Efficient technologies and supportive policy could hold demand close to current levels, but the report states that climate-driven increases in cooling, water pumping and other energy services may offset efficiency gains and constrain demand-side emissions reductions. At the same time, heat undermines low-carbon energy systems themselves. Prolonged droughts sharply reduce hydropower generation, heatwaves lower solar photovoltaic efficiency, and changes in wind patterns make output harder to predict. The report also describes a reinforcing feedback in which drought raises fossil fuel reliance through reduced hydropower and increased cooling demand.
Urban habitability. The urban heat island effect combined with rising mean and extreme temperatures is projected to worsen overheating in cities globally, with negative consequences for health, comfort and productivity. Relative to a 1.5°C pathway, a 3°C scenario could extend heatwave duration by an average of 28 per cent, raise annual heatwave frequency by 25 per cent, and increase cooling energy demand. At 1.8°C of warming, 76 per cent of urban tree species will be at risk by 2050 from projected changes in mean annual temperature and precipitation, hampering the potential of this kind of nature-based solution at the very time it is most required.
Health. Heat-related mortality can rise sharply. Events of a scale expected only once every 100 years under the climate of 2000 now occur every 10 to 20 years under that of 2020. As warming reaches 1.5°C to 2°C, summers in which deaths rise steeply are expected to become commonplace without adaptation. The report notes that indirect impacts, especially food insecurity driving malnutrition and diet-related disease, may exceed direct heat risks.
Conditions of work. Threats from heat and humidity are experienced differently according to occupation, working conditions, income, social roles and gender. Drawing on the World Bank's country reports, the report cites an estimate that at 3°C above 1990 levels, labour capacity could fall by about 11 per cent for manual labourers against less than 1 per cent for office workers. It notes that this estimate uses a 1990 baseline not directly comparable with the pre-industrial baseline of the Paris Agreement. The same rise in temperature lands more than ten times harder on one occupation than on another.
Government institutions and their legitimacy. The report states that national and local institutions which fail during the peak may not fully recover even if climatic conditions later improve. Even if temperatures come down, broken institutions do not come back. Unequal impacts erode public trust, and legitimacy is challenged where long-term inequities and abandoned areas are not explicitly addressed. Against that, the report notes that effective national and subnational institutions can substantially reduce mortality from extreme events regardless of income level.
Beyond these, the report identifies education, the ability to decide for oneself whether and when to move, intergenerational equity and the capacity of humanitarian systems, among others, as things that can be degraded, along with the burden of having to adapt over and over again.
One further structural warning deserves mention. Poorly designed adaptation can reinforce existing social, economic, infrastructural, technological and ecological risks and vulnerabilities across sectors, regions and timeframes. The report's example is hard coastal defence, which can protect high-value assets while shifting risk to lower-income communities and foreclosing longer-term options such as managed retreat. Adaptation that is optimal for a single asset can be a net loss at the scale of the locality.
05
Translated into buildings
Rendered at the scale of a single building, these forms of erosion become questions of specification and operation.
Envelope and solar control move from energy saving to a condition of continued occupancy. Cooling demand could more than treble, and efficiency gains may be offset by that increase. Together these change what insulation and shading are for. Utility bill savings are a weak account of the benefit. When cooling becomes a condition of habitability, what matters is how long a building can hold interior conditions without depending on that electricity. That length of time decides whether people can stay in the building and whether a business can keep running.
On-site solar becomes something whose resilience to climate must itself be tested. Heatwaves lower photovoltaic efficiency, and renewable energy systems are described as increasingly exposed to climate variability and extremes, which makes their output harder to predict. A decarbonisation plan premised on on-site generation may lose output at precisely the moment it is needed. Plans need testing against output under extreme conditions, not against expected annual yield.
Separate the electricity, water and telecommunications systems so that a power failure does not stop water and communications. During the peak, the report finds a high likelihood of rapid cross-sector cascading failures spanning energy, water, information and communications technology, and health, arising from compound extreme weather events. Its adaptation principles call for decoupling these systems where possible and containing failures locally. Backup power, rainwater harvesting and a second communications line are not additional resilience spending. A power cut stops the pumps, the pumps stop the water supply, and the loss of water closes the building. These are the measures that break that sequence part-way along.
Green infrastructure must answer for its own resilience. That finding on urban tree species does not argue against planting. It means species selection and irrigation design come before the question of whether to plant.
Stop trimming design margin. The report's principle for the exceedance stage is design margin extension, not optimisation: prioritise durability and safety margin, and accept inefficiency in exchange for being able to keep the building in service. It also calls for shifting from historical risk planning to forward-looking stress testing, and adapting before failures become too frequent. This sits in direct conflict with the instincts of value engineering.
Build redundancy into materials procurement. The report notes that climate impacts and geopolitical instability are likely to disrupt supply chains for construction materials and energy, and that increasing supplier redundancy across regions and climate zones provides greater resilience.
06
Asset management takes different action at each stage
Two pathways. The solid line assumes deep mitigation is delivered, and shows the exceedance, peak and decline stages with the human action that defines each. The dashed line is current policies: no peak arrives this century, and warming in 2100 is projected at around 2.6°C in central estimates. Years and temperatures are indicative, based on the report's statements that warming crosses 1.5°C within the next decade, that global net-zero CO2 falls in the 2050s, and that even the most optimistic scenario peaks at around 1.8°C. This is a schematic: the shape of the curves and the timing are not intended as precise values. UNEP sets out four stages — exceedance, peak, decline and stabilisation — of which this article treats three. Source: UNEP, Limiting Overshoot (2026).
A word first on what the two lines in the schematic mean. The solid line is the shape that appears only if the world reaches net zero in the 2050s and then goes further, into net negative. It is what should be achieved, not a forecast of what will happen. The dashed line is what is currently in prospect under existing policies. The report states plainly that even the solid pathway is by no means an acceptable or preferred one; it is simply the best remaining option.
A peak, therefore, is not guaranteed. If it does not arrive, the exceedance stage simply does not end, and the peak and decline principles never come into play. Of the three stages, only the exceedance principles are certain to be needed. If the question is where to start, it is there.
The report contains a table setting out adaptation goals and principles for infrastructure and the built environment at each stage of overshoot. For asset management, this is its most usable section. Three of its principles are set out below as who does what and when, with indicative years matching the schematic above.
Exceedance stage (from now until a peak arrives). Owners and asset managers fix, now, the numbers that trigger the work. Map explicitly which assets are exposed to which hazard, identify flood levels and temperature thresholds, and make those the trigger for adaptation investment. That is the report's principle. Rather than relitigating the case for a retrofit each year, decide beforehand what has to reach what level before work begins. Once that is settled, the decision load falls and the work does not start late. The same stage also calls for keeping room for retreat and change of use, because by the time a crisis forces the question of abandonment, the useful options are gone.
Peak stage (if net zero is reached, around the 2050s). The owner decides which holdings to protect and which to release. Avoid abandoning an asset in practice by simply not attending to it, and decide explicitly which assets are to be protected, which are to be allowed to degrade, and which are to be released. That is the report's principle. It names three things to avoid: assets that lose their value and can be neither sold nor converted, money spent with no prospect of recovery, and long-term liabilities that tie up future investment capacity. Holding on without deciding turns out to be the most expensive option.
Decline stage (if the world goes net negative, 2070s onward). Owners and designers review, one by one, every emergency measure taken at the peak. Parts rebuilt in haste after damage, and stopgap repairs, tend to become permanent in practice; left alone, they lock vulnerabilities into the next generation of assets. The report's principle is to inspect every decision and stopgap made during the peak and settle each one individually: reverse it, rebuild it to a durable specification, relocate it, replace it or upgrade it. Pre-overshoot conditions no longer serve as a reference, so design standards, safety margins and land use all have to be redrawn on the assumption of a changed climate.
Stage
Adaptation goal
Principles
Exceedance
Buy time by absorbing shocks and constraining risks without foreclosing future adaptation pathways
Anticipatory adaptation over reactive response / design margin extension, not optimisation / no-regret and reversible measures first / early identification of irreversible thresholds / keep options open for retreat and reconfiguration
Peak and plateau
Normalise operation beyond continued exceedance of design limits, preventing cascading failures and avoiding locked-in irreversible losses
Safety over performance / active triage of assets / operate for endurance, not recovery / minimise additional irreversible damage / reduce system coupling / legitimate emergency governance with safeguards
Decline
Move from reactive endurance to long-term stewardship within a permanently altered climate envelope
Rebaseline, do not restore / govern irreversible loss explicitly / replace emergency fixes with durable solutions / adjust service provision to environmental limits / institutionalise adaptation and learning / equitable governance of irreversible losses
Source: summary of Table 4.1 (infrastructure and the built environment) in UNEP, Limiting Overshoot.
07
The return on the spending is the loss avoided
Start with the financial side. As climate risks intensify, the report states, the insurability of infrastructure assets could decline. Combined with rising sovereign debt, that limits investment in resilience upgrades and in repurposing. The result is a threat to the economic viability of roads, rail and transmission lines, of housing, and of utilities. Insurers may withdraw from high-risk regions before a peak is reached, shifting costs to governments and households and eroding the property values and tax base on which local adaptation depends. The means of funding adaptation are consumed by the very conditions that make adaptation necessary.
The report also identifies a feedback loop in which overshoot prolongs itself. During the overshoot period the need for emergency response and reconstruction increases. If meeting that need diverts fiscal and institutional resources away from sustained mitigation and adaptation, there is a risk of prolonging the severity and duration of the overshoot itself. Conversely, adaptation that reduces vulnerability also reduces future demands on reactive and emergency spending, and preserves the fiscal capacity needed to sustain the steep reductions through net zero and into net negative that bring temperatures back down.
Adaptation investment is not disaster expenditure. It is investment in preserving the capacity to keep decarbonising.
Where the money spent on adaptation and decarbonisation comes back from has to be taken in two separate cases.
The first is the contribution to the planet. A low-carbon building does contribute to limiting global warming, but the contribution of any single asset is vanishingly small. The report gives a best estimate that every 220 billion tonnes of carbon dioxide produces 0.1°C of global warming. No account of one building's benefit survives at that order of magnitude. The harm avoided disperses thinly across the world, so it is impossible to say who was spared, and what cannot be identified cannot be priced.
The second is the benefit that lands directly on the locality where the building stands and on the people who use it. Here it is clear who is spared. In a world where electricity demand for cooling rises steeply, a building with a high-performing envelope lowers occupants' energy bills, holds interior conditions through a heatwave, and protects the continuity of both occupancy and employment. A building that contains cascading failure locally reduces a tenant's days of interrupted operation. In each case it is clear who benefits, and the effect returns through what tenants can afford in rent, occupancy, days of trading lost, and how readily the asset can be insured.
What comes back to investors, in our view, is not a subsidy or concession granted from outside but avoided loss and income that was not forfeited. The routes we consider realistic are four: underwriting terms, lending terms, reduced emergency expenditure by municipalities, and tenant demand.
The figures needed to support these decisions, however, remain thin. The report notes that a recent systematic global stocktake of adaptation research finds adaptation to be largely fragmented, local and incremental rather than transformational, and that an entrenched imbalance between mitigation and adaptation finance leaves adaptation chronically under-resourced even in the absence of crisis. It adds that current monitoring, evaluation and learning frameworks are poor at capturing cases where adaptation backfired and poor at supporting learning from them, and that this limitation becomes more serious under overshoot conditions. A measure whose effect has not been quantified becomes a proposal in which only the cost side is certain. When such proposals fail to clear internal approval, the cause is not the presenter's skill. It is that the effect of adaptation has not yet been put into numbers. Producing those numbers is the work that comes next.
08
The cause is human activity, and its conditions are fixed by buildings
A closing word on where responsibility sits.
The report records that in its 2025 advisory opinion the International Court of Justice determined 1.5°C to be the agreed primary temperature goal of the Paris Agreement parties, and that all States must exercise stringent due diligence to regulate greenhouse gas emissions including by private actors. The standard is not fixed. Due diligence evolves with scientific knowledge and with the probability and gravity of foreseeable harm, and as exceedance increases both, due diligence becomes more demanding. Overshoot does not lighten the obligation; it tightens it.
Buildings are where that obligation lands. The report states that infrastructure and buildings have long lifespans, locking in emissions and resilience outcomes for decades or centuries to come. Buildings do not themselves cause climate change. Human activity does, and buildings set the conditions under which that activity takes place, then hold those conditions in place for decades. Today's design decisions set the range of options available to the next generation.
That responsibility is proportionate to the size of the sector, as set out in the Global Status Report for Buildings and Construction 2025-2026, published by UNEP and the Global Alliance for Buildings and Construction in May 2026. Inger Andersen, Executive Director of UNEP, stated that buildings can either lock in climate risks or deliver safer, healthier and more affordable living conditions. The report expects around half of the buildings that will exist in 2050 to be built or renovated between now and then.
The scale of buildings and construction
11–13%of global GDP, from buildings and construction
about 9%of the world's workforce, employed by the sector
37%of global carbon dioxide emissions, from building operation and construction combined
Whether new build or retrofit, the envelope, the plant and the use of the site are settled once the work is complete, and are not easily changed for the decades that follow. Half of the buildings standing in 2050 are expected to go through that work between now and then. The performance written into today's drawings is the performance those buildings will have in 2050.
Three proposals.
Draw the route map for one building. Among water, energy, urban habitability, health and conditions of work, select the items that bear on that asset and draw a line from the change in climate through to what happens to its occupants and workers. Wherever the line cannot be drawn is where the next data collection begins.
Fix the numbers that trigger the work. What inundation depth, what run of extreme heat days, what duration of power loss starts which job. Deciding the conditions in advance beats relitigating a retrofit every year, and the work does not start late. Begin the discussion of retreat and change of use before a crisis forces it.
Pool the recording format across the industry. The record does not accumulate because no shared format exists for what to record and at what level of detail. The information sits in different hands: those preparing technical due diligence reports, appraisers, asset managers, lenders, insurers, and the operators who run the facilities.
Carbon that was not cut turns into a burden on society, and comes to rest on buildings.
GOYOH Inc. provides a platform that connects local regeneration with investment and lending, with social impact as the common axis. Our role is to generate social value from buildings and, by connecting neighbourhoods and the people who use them with investment and finance, to keep that value circulating and widening.