The atmosphere is a ledger, and it does not round numbers. Every year we post more carbon to one side than the planet can take off the other. This page is not a report — it is a reading, taken in public, so that anyone doing sustainability work in a supply chain can benchmark their own common sense against the world's actual arithmetic.
大气是一本账，它不会四舍五入。每一年，我们记入的碳，都多过地球能够划走的碳。这一页不是报告，而是一次公开的读数——让任何在供应链里做可持续的人，都可以拿自己的常识，去对一对这个世界真实的算术。
问题不是"我们做得够不够好"，而是"我们盖住的，是不是最该盖的那一块"。
Humanity now releases roughly 41.6 billion tonnes of CO₂ a year — about 37.4 Gt from fossil fuels and cement, plus around 4.2 Gt from land-use change. Counting all greenhouse gases, the total was 57.1 GtCO₂e in 2023, a record, and still rising 1.3% year on year.
人类每年释放约 416 亿吨二氧化碳：化石燃料与水泥约 374 亿吨，土地利用变化约 42 亿吨。若把所有温室气体算在一起，2023 年为 571 亿吨二氧化碳当量，创下纪录，并仍以每年 1.3% 增长。
Almost half of what we emit — the airborne fraction, around 45–50% — simply stays up there, for centuries. That surplus is the whole story: not the emissions alone, but the gap between emission and absorption. Concentration rose about 3.5 ppm in 2024, the fastest jump ever recorded.
我们排放的将近一半（"大气存留比例"约 45–50%）会在空中停留数百年。真正的故事不是排放本身，而是排放与吸收之间的差额。2024 年浓度上升约 3.5 ppm，是有记录以来最快的一次跃升。
Sources: NOAA Global Monitoring Lab · Global Carbon Project, Global Carbon Budget 2024 · UNEP Emissions Gap Report 2024. See full reference list ↓
Share of global greenhouse gas emissions by sector. Note what this really says: 73% of all emissions are energy — burning something to make heat, motion or electricity. Climate change is, first and last, an energy problem wearing a hundred different industrial costumes.
全球温室气体排放的部门占比。请注意它真正在说什么：73% 的排放来自能源——为了热、动力或电力而燃烧某种东西。气候问题归根到底是一个能源问题，只是穿了上百种工业的外衣。
Inside those bars: road transport 11.9% · iron & steel 7.2% · livestock & manure 5.8% · cement 3.0% · aviation 1.9% · shipping 1.7%.
柱子内部：公路运输 11.9% · 钢铁 7.2% · 畜牧与粪肥 5.8% · 水泥 3.0% · 航空 1.9% · 航运 1.7%。
Source: Climate Watch / Our World in Data, sectoral shares of 49.4 GtCO₂e — the most complete cross-checked breakdown available; totals have grown since, shares have moved only slightly.
For sixty years the ocean and the land did us an enormous unpaid favour: they absorbed more than half of everything we burned. That favour is conditional. A warmer ocean dissolves less CO₂. Forests stressed by drought, fire and clearing flip from sink to source — in the 2023–24 El Niño year the land sink weakened sharply and atmospheric CO₂ leapt at a record rate even though emissions grew only slightly.
So the imbalance has two engines, not one. We push harder on the emitting side, and the absorbing side loses strength at the same time. Any strategy that only counts tonnes emitted, and never counts sink capacity protected, is reading half the ledger.
六十年来，海洋与陆地为我们做了一件巨大的无偿工作：吸走了我们燃烧掉的一半以上。但这份好意是有条件的。海水变暖，能溶解的二氧化碳就变少；森林在干旱、火灾与砍伐下承压，会从"汇"翻转为"源"。在 2023–24 厄尔尼诺年，陆地碳汇明显走弱，即使排放增幅有限，大气浓度仍以创纪录的速度跃升。
所以失衡有两台发动机，而不是一台：我们在排放端更用力，吸收端同时在变弱。任何只统计"排了多少吨"、却从不统计"保住了多少吸收能力"的策略，都只读了半本账。
另外三个放大器：甲烷（约占迄今升温的三分之一，但在大气中只停留十余年，因此减排见效最快）；燃烧减少带来的气溶胶降温效应正在消退；以及冰、永冻土与森林中的自我强化反馈。
We are not writing a report. We are benchmarking our common sense.
UNEP's 2024 assessment finds the world could cut 31 GtCO₂e a year by 2030 using measures costing under US$200/t — enough, on paper, to put 1.5°C back in reach. The percentages below are shares of that available 2030 potential, or of global emissions where a sector share is the honest number. Ranked, because size matters more than novelty.
联合国环境署 2024 年评估认为，全球到 2030 年可用低于 200 美元/吨的措施，每年减少 310 亿吨二氧化碳当量——在纸面上足以让 1.5°C 重回视野。下列百分比，指的是在这份 2030 年可实现潜力中的份额，或该部门在全球排放中的占比。之所以排序，是因为"体量"比"新意"更重要。
The single largest block. Solar PV and wind together could deliver 27% of all available 2030 reductions and 38% by 2035 — because they are already the cheapest new power in most markets.
最大的一块。光伏与风电合计可贡献 2030 年全部可实现减排的 27%、2035 年的 38%——因为在多数市场，它们已经是最便宜的新增电源。
Forest-related action — halting clearing, restoring degraded land, better management — offers around a fifth of the total potential, and it defends the sink at the same time. Two sides of the ledger, one intervention.
与森林相关的行动——停止砍伐、恢复退化土地、改善经营——约占总潜力的五分之一，同时还在守护碳汇。一个动作，作用于账本的两边。
Oil, gas and coal leak roughly 80 Mt of methane a year. The IEA finds about 70% is technically avoidable with existing kit, and around 40% at no net cost. Methane leaves the air in ~12 years, so this is the fastest brake we own.
石油、天然气与煤炭每年泄漏约 8000 万吨甲烷。IEA 认为其中约 70% 用现有技术即可避免，约 40% 甚至零净成本。甲烷在大气中仅停留约 12 年——这是我们手上最快见效的刹车。
Around one in five new cars sold worldwide is now electric. The gain is only real where the grid is cleaning up at the same time — the two levers must be pulled together, or an EV just moves the tailpipe to a coal plant.
全球每卖出五辆新车，就有约一辆是电动车。但只有在电网同步变清洁时，减排才真正兑现——两个杠杆必须同时拉，否则电动车只是把排气管搬到了燃煤电厂。
Doubling the global rate of energy-efficiency improvement to about 4% a year is, in IEA pathways, one of the largest single contributors to emission reductions before 2030. The cheapest tonne is the one never needed.
把全球能效改善速度提高一倍至每年约 4%，在 IEA 的路径中是 2030 年前贡献最大的单项之一。最便宜的一吨碳，是从未被需要的那一吨。
Steel ≈2.6 Gt and cement ≈2.4 Gt of CO₂ a year. Cement's problem isn't only fuel — over half is chemistry, released when limestone becomes clinker. Hydrogen DRI, clinker substitution, and simply using less material through design.
钢铁约 26 亿吨、水泥约 24 亿吨二氧化碳/年。水泥的难点不只是燃料——一半以上来自化学反应：石灰石变熟料时释放的碳。氢基直接还原铁、熟料替代，以及通过设计"少用材料"。
About a fifth of food available to consumers is wasted; food systems as a whole are roughly a third of global emissions. Nothing else on this list is so widely distributed across households, retailers and logistics at once.
消费端约有五分之一的食物被浪费；整个食物系统约占全球排放的三分之一。这份清单上，没有第二项像它这样，同时分散在家庭、零售与物流之中。
Coal is still the single largest source of energy-related CO₂ and the largest source of global electricity. Every year a new plant is built, thirty years of emissions are booked in advance.
煤炭仍是能源相关二氧化碳的最大单一来源，也是全球电力的最大来源。每建一座新电厂，就等于提前预订了三十年的排放。
Full implementation of the Kigali Amendment on HFCs could avoid up to half a degree of warming this century — while cooling demand itself is set to soar. Efficiency and refrigerant choice are the same decision, made once, in a design office.
全面落实关于氢氟碳化物的《基加利修正案》，本世纪可避免最多 0.5°C 升温——而制冷需求本身正在飙升。能效与制冷剂选择，其实是设计阶段做的同一个决定。
The IEA counts millions of kilometres of grid needing to be added or refurbished by 2040. This lever emits nothing itself — it decides whether the top of the list can be built at all. Queue length, not cost, is now the binding constraint in many markets.
IEA 估算，到 2040 年需新建或改造数百万公里电网。这个杠杆本身不减排——它决定了清单顶端的那些事能不能落地。在许多市场，真正的瓶颈已不是成本，而是并网排队。
Two lists, and they are not the same list. Strategy that only chases today's biggest emitters will keep being surprised by what tomorrow's supply chain quietly grew.
两份清单，而且它们并不相同。只盯着今天最大排放源的策略，会不断被明天悄悄长起来的供应链打个措手不及。
CDP's supply-chain analysis finds that, for a typical company, emissions in the value chain are on average about 11.4 times its own operational emissions. That single ratio should reorganise most sustainability job descriptions. The site you control is not where the carbon is. The specification you write, the supplier you qualify, the material you designed in three years ago — that is where it is.
Which reframes the honest worry behind this page: are we covering the best of what we can cover? Often the answer is no — not because teams are lazy, but because measurement follows ownership, and carbon follows purchasing. Coverage improves when procurement criteria change, not when the report gets longer.
So the practical benchmark is not "how many tonnes did we report" but: what share of spend is covered by supplier-specific data rather than industry averages? How many of our top-20 suppliers have validated targets? Did we design a kilogram of steel or a kilowatt-hour of compute out of the product this year?
CDP 的供应链分析发现，对一家典型企业而言，价值链排放平均约为其自身运营排放的 11.4 倍。仅这一个比值，就应当重写大多数可持续岗位的职责说明。碳并不在你所控制的那个厂区里，而在你写下的规格书、你认证的供应商、你三年前定下的那个材料里。
这也让这一页背后那个诚实的疑问变得清晰：我们盖住的，是我们能盖住的最好的那一块吗？ 常常答案是"不是"——不是因为团队不努力，而是因为"测量跟着所有权走，碳跟着采购走"。覆盖度的提升，发生在采购准则改变的时刻，而不是报告变长的时刻。
因此，真正的基准不是"我们报了多少吨"，而是：有多少采购额是用供应商实测数据、而非行业平均值覆盖的？前二十大供应商中，有多少家有经过验证的减排目标？今年，我们有没有从产品里"设计掉"一公斤钢，或一度算力用电？
If, reading the numbers above, a question arrived — that was the best moment on this page. Here are the ones that arrived while writing it.
如果读到上面的数字时，你心里冒出了一个问题——那就是这一页最好的时刻。以下是写这一页时冒出来的那些。
Are we reducing carbon, or relocating it — offshore, downstream, into someone else's scope 1?
我们是在减碳，还是在搬碳——搬到海外、搬到下游、搬进别人的范围一？
If 73% of emissions are energy, why does most of our reporting effort go to the other 27%?
如果 73% 的排放来自能源，为什么我们大部分的报告精力，花在了剩下的 27% 上？
Does our footprint improve when we sell less — and if so, is that a strategy or an accident?
我们的碳足迹是不是"卖得少就变好"？如果是，那是策略，还是意外？
Who in our supply chain has no capital to decarbonise — and what would it cost us to lend it?
供应链里谁根本没有减碳的资本？我们借给他，要花多少？
The sink is weakening. Does any line in our plan protect absorption, or do we only count emission?
碳汇在变弱。我们的计划里，有没有一行是在保护"吸收"，还是只在数"排放"？
What did we design out this year — a material, a trip, a kilowatt-hour? Name it, or it didn't happen.
今年我们"设计掉"了什么——一种材料、一趟运输、一度电？说得出名字，才算发生过。
Carbon watching is a practice, not a conclusion: read the reading, ask the question, find the room, act, then benchmark again — and share it.
观碳是一种练习，不是一个结论：读数、提问、找到行动空间、行动，然后再次校准——并且分享出去。
Every figure on this page comes from a public dataset. Where sources differ in year or boundary, the text says so. If a number here is stale or wrong, the right response is not to trust the page — it is to open the link.
本页每一个数字都来自公开数据。若不同来源在年份或核算边界上有差异，正文已注明。如果这里的数字过时或有误，正确的反应不是相信这一页，而是打开链接。