Global energy dossier · 2020 → 2050 · data station
全球能源档案 · 2020 → 2050 · 数据站
Between 2020 and 2025 the world's oil and gas companies pumped more, promised more, and spent less than five cents of every capital dollar on the future they keep describing. This station lays out the roadmap — global, by region, along the whole value chain — in numbers you can check.
2020 到 2025 年，全球油气公司的产量创下新高，承诺越来越漂亮，而真正投向"未来"的资本支出不足 5%。这个站点把路线图摊开来看——全球、分大洲、贯穿整条产业链——全部用可核对的数据说话。
The pledges point to 2050. The drilling schedules point to 2035. That gap is the whole story.
承诺指向 2050，钻井计划指向 2035。这道缝隙，就是全部故事。
Global oil supply in 2025 — an all-time high, above pre-COVID 2019.
2025 年全球石油供应，历史最高，已超 2019 年疫情前水平。
Share of oil & gas industry capital spending that goes to clean energy.
油气行业资本支出中真正投向清洁能源的比例。
Bank financing arranged for fossil fuels in 2024 — up on the year before.
2024 年银行为化石燃料安排的融资，比前一年还多。
Assessed oil & gas companies whose capital plans align with 1.5°C.
被评估的油气公司中，资本计划与 1.5°C 对齐的数量。
The pandemic dip was mistaken for a turning point. It was a pothole. Oil supply has added roughly ten million barrels a day since the 2020 trough; natural gas set fresh records in 2024 and again in 2025.
疫情造成的下跌曾被误读为转折点，其实只是一个坑。自 2020 年低谷以来，石油供应回升约每日一千万桶；天然气在 2024 年、2025 年连续刷新纪录。
million barrels / day · scale starts at 85 · 2025 estimated
百万桶/日 · 刻度自 85 起 · 2025 为估算
billion cubic metres · scale starts at 3,600 · 2025 estimated
十亿立方米 · 刻度自 3,600 起 · 2025 为估算
Verdict: increasing, on both fuels, through 2025.
结论：截至 2025 年，两种燃料都在增产。
There is no single peak date — there are competing futures, each attached to a balance sheet. Mainstream forecasters see demand flattening in the late 2020s. Producer forecasts see no peak at all. The 1.5°C pathway needs the decline to have begun already.
并不存在唯一的峰值年份，而是几种互相竞争的未来，每一种都挂在一份资产负债表上。主流机构认为需求在 2020 年代末走平；产油方的预测里根本没有峰值；而 1.5°C 路径要求下降从现在就开始。
Consensus reading: demand plateau 2027–2032, supply peak nowhere in corporate plans.
综合判断：需求平台期约在 2027–2032，而公司计划里看不到供应峰值。
Decline is regional, not global. Europe is genuinely shrinking; the Americas and the Gulf are absorbing the difference. Every barrel Europe stops producing is being re-sanctioned somewhere with lower costs and looser rules.
下降是区域性的，不是全球性的。欧洲在真实收缩，而美洲与海湾正把缺口补回来。欧洲少产的每一桶油，都在成本更低、规则更松的地方被重新批准。
US crude at a record ~13.5 mb/d in 2025, but shale growth is decelerating. Canada ~5.6 mb/d after Trans Mountain. LNG export capacity roughly doubles by 2030.
2025 年美国原油约 1,350 万桶/日创纪录，但页岩增速放缓。加拿大在跨山管道后约 560 万桶/日。LNG 出口能力到 2030 年约翻倍。
Liquids plateau ~2027–2030 · gas keeps rising
液体燃料约 2027–2030 走平 · 天然气继续上升
Saudi Arabia holds 12 mb/d capacity (the 13 mb/d expansion was cancelled in 2024). UAE to 5 mb/d by 2027. Iraq targets 6 mb/d. Qatar lifts LNG from 77 to 142 mtpa by 2030.
沙特维持 1,200 万桶/日产能（1,300 万桶扩产于 2024 年取消）。阿联酋 2027 年达 500 万桶/日。伊拉克目标 600 万桶/日。卡塔尔 LNG 从 7,700 万吨提至 2030 年 1.42 亿吨。
Capacity-led growth · no peak before 2035
以产能扩张为主 · 2035 年前无峰值
The fastest growth on Earth. Guyana from 0.65 to ~1.7 mb/d by 2030; Brazil's pre-salt toward 4.5 mb/d; Argentina's Vaca Muerta from 0.8 to ~1.5 mb/d.
全球增速最快。圭亚那由 65 万桶/日增至 2030 年约 170 万桶/日；巴西盐下层趋向 450 万桶/日；阿根廷 Vaca Muerta 由 80 万增至约 150 万桶/日。
Peak in the 2030s at the earliest
最早也要到 2030 年代才见峰值
Structural North Sea decline of 4–6% a year. Norway peaks around 2025. UK licences are being fought in court, and refining capacity is closing.
北海结构性下降，年均 4–6%。挪威约在 2025 年见顶。英国新许可在法庭上被挑战，炼能持续关停。
Peak passed decades ago · fastest real decline
峰值已过数十年 · 真实下降最快
Mature decline in Nigeria and Angola, offset by frontier plays: Namibia's Orange Basin, Senegal–Mauritania, Uganda's EACOP, Mozambique LNG. Flaring intensity remains the world's worst.
尼日利亚、安哥拉成熟油田衰退，被新前沿抵消：纳米比亚奥兰治盆地、塞内加尔—毛里塔尼亚、乌干达 EACOP、莫桑比克 LNG。放空燃烧强度仍居世界之首。
Late-2030s peak if frontiers proceed
若新前沿推进，峰值约在 2030 年代末
China's crude output is flat near 4.3 mb/d while gas, storage and offshore expand — and its oil demand may already have peaked as EVs take a third of new car sales. India is now the fastest-growing demand centre. Australian LNG plateaus.
中国原油产量稳定在约 430 万桶/日，天然气、储气与海上业务扩张；随着电动车占新车销量三分之一以上，石油需求可能已见顶。印度成为增速最快的需求中心。澳洲 LNG 走平。
Supply flat · demand pivot is the real story
供应走平 · 真正的变化在需求端转向
Sanctions and OPEC+ quotas cap oil near 9–9.5 mb/d. Pipeline gas to Europe is down roughly 80% since 2021; Arctic LNG projects are stalled for want of technology and buyers.
制裁与 OPEC+ 配额将石油压在约 900–950 万桶/日。对欧管道气自 2021 年下降约 80%；北极 LNG 因技术与买家缺位而停滞。
Peak passed · decline is geopolitical, not chosen
峰值已过 · 下降源于地缘政治，而非主动选择
Oil & gas operations alone emit about 5.1 Gt CO₂-equivalent a year — roughly 15% of energy-related emissions — before a single product is burned. Add combustion and it is over half.
仅油气生产环节每年排放约 51 亿吨二氧化碳当量，约占能源相关排放的 15%——这还没算产品燃烧。计入燃烧后超过一半。
Flaring hit ~151 bcm in 2024, a ten-year high
2024 年放空燃烧约 1,510 亿立方米，十年新高
Pressure lands almost entirely on the eight or nine listed producers everyone can name. The rest of the chain — services, traders, pipelines, petrochemicals, insurers — carries comparable emissions with a fraction of the scrutiny. Dots below show how much daylight each link actually gets.
压力几乎全部压在人人叫得出名字的那八九家上市公司身上。产业链的其余环节——油服、贸易商、管道、石化、保险——承担着可比的排放，却只受到极小部分的监督。下面的圆点表示每个环节被"照见"的程度。
SLB, Halliburton, Baker Hughes, Transocean. Enable nearly every new well; none carry a Scope 3 target.
斯伦贝谢、哈里伯顿、贝克休斯、越洋钻探。几乎每口新井都靠它们，却无一设定范围三目标。
●●○ scrutiny监督度
Over 60% of world output is state-owned. These are the decisive actors and the least exposed to shareholder votes.
全球产量六成以上为国有。它们是决定性角色，却最不受股东投票约束。
●●● scrutiny监督度
Enbridge, TC Energy, Cheniere, QatarEnergy trains. 20–40 year assets locked in this decade decide the 2050s.
恩桥、TC 能源、Cheniere、卡塔尔 LNG 生产线。本十年锁定的 20–40 年资产，决定 2050 年代。
●●○ scrutiny监督度
Jamnagar, Sinopec, Marathon. Closing in Europe, expanding in Asia and the Gulf — capacity migrates rather than retires.
贾姆纳格尔、中石化、马拉松。欧洲关停、亚洲与海湾扩张——产能是在迁移，而非退役。
●○○ scrutiny监督度
SABIC, INEOS, Dow, Braskem. Oil's plan B: the single largest source of oil demand growth to 2030.
沙特基础工业、英力士、陶氏、Braskem。石油的 B 计划：到 2030 年最大的需求增长来源。
●○○ scrutiny监督度
Vitol, Trafigura, Glencore, Mercuria move a large share of seaborne crude with minimal climate obligation.
维多、托克、嘉能可、摩科瑞掌控大量海运原油，气候义务却极少。
●○○ scrutiny监督度
Forecourts, aviation and shipping fuel contracts. Where EV and e-fuel policy bites first and hardest.
加油站、航空与航运燃料合约。电动化与电制燃料政策最先、最狠地在此发力。
●●○ scrutiny监督度
Banks, insurers, export credit agencies, sovereign funds. Nothing gets drilled that cannot be financed and underwritten.
银行、保险公司、出口信贷机构、主权基金。融不到钱、投不到保，就打不成井。
●●● the real lever真正的杠杆
The honest answer to "is it net positive yet": no. Record cash from 2022–2024 went overwhelmingly to shareholders and to more oil and gas. Clean energy is a rounding error on the industry's own capital plans, and the industry supplies about 1% of the world's clean energy investment.
对"是否已经转为净正向"这个问题，诚实的答案是：还没有。2022–2024 年的创纪录现金，绝大部分流向股东和更多油气项目。清洁能源在行业自己的资本计划里只是一个尾数——全球清洁能源投资中，油气行业只贡献约 1%。
Shareholders vs. the transition
股东回报 vs. 能源转型
Since 2021 the five Western majors have returned well over $300bn in dividends and buybacks — roughly eight times their combined low-carbon spending over the same period.
自 2021 年以来，五大西方巨头以分红与回购形式返还股东逾 3,000 亿美元，约为同期低碳投入总额的八倍。
Bank financing ratio · low-carbon per $1 of fossil
银行融资比 · 每 1 美元化石对应的低碳投入
Banks arranged about $869bn for fossil fuels in 2024 — more than the year before, and about $7tn since the Paris Agreement. The green:red ratio has to quadruple this decade.
2024 年银行为化石燃料安排约 8,690 亿美元融资，比前一年更多；《巴黎协定》以来累计约 7 万亿美元。绿红比必须在本十年翻两番。
Verdict
结论
On public disclosures, no listed major spends more on decarbonisation than it hands back to shareholders. "Give-back positive" is still a slogan, not an accounting fact.
依公开披露，没有一家上市巨头在脱碳上的支出超过其返还给股东的金额。"回馈为正"目前仍是口号，而非会计事实。
Very little goes to nature. Most "climate capex" is engineering inside the fence line: carbon capture at their own plants, hydrogen for their own refineries, and CO₂ that frequently ends up pushing more oil out of old reservoirs.
投给自然的极少。多数"气候资本支出"都是围墙之内的工程：自家装置的碳捕集、自家炼厂的氢气，而捕来的二氧化碳往往被重新注入老油藏，去驱出更多的油。
Roughly 50–55 Mt CO₂ a year of operating capacity worldwide. The 1.5°C pathway asks for about 1,000 Mt by 2030 — a twentyfold build in five years.
全球在运产能约每年 5,000–5,500 万吨。1.5°C 路径要求 2030 年达到约 10 亿吨——五年内扩大二十倍。
Status: far behind
状态：严重落后
Historically the large majority of captured CO₂ has been used for enhanced oil recovery. Removal that funds extraction is not removal.
历史上绝大部分被捕集的二氧化碳用于提高石油采收率。为开采买单的"移除"，不是移除。
Status: accounting risk
状态：核算风险
Occidental's Stratos, the largest DAC plant, is designed for 0.5 Mt a year — about 0.001% of annual global emissions, at premium cost.
西方石油的 Stratos 是全球最大 DAC，设计年捕集 50 万吨——约相当于全球年排放的 0.001%，且成本高昂。
Status: real, but tiny
状态：真实，但极小
Retreating after integrity scandals: Shell shelved its $100m-a-year offset programme in 2023. Industry reforestation pledges total a fraction of a percent of their emissions.
在信用度丑闻后退潮：壳牌 2023 年搁置了每年 1 亿美元的抵消计划。行业造林承诺加总也只占其排放的极小比例。
Status: shrinking, not scaling
状态：在收缩，而非扩张
Cutting fossil methane 75% by 2030 would cost on the order of $170bn — under 5% of the industry's 2022 net income — and is the fastest lever on near-term warming.
到 2030 年将化石甲烷减少 75%，成本约 1,700 亿美元——不到该行业 2022 年净利润的 5%——却是抑制近期升温最快的杠杆。
Status: pledged at COP28, under-delivered
状态：COP28 已承诺，兑现不足
Around 151 bcm of gas burned at the wellhead in 2024 — a ten-year high, despite a "zero routine flaring by 2030" pledge signed by more than fifty companies.
2024 年在井口烧掉约 1,510 亿立方米天然气，十年新高——尽管五十多家公司签署了"2030 年零常规放空燃烧"承诺。
Status: moving backwards
状态：在倒退
Commitments for 2030, 2040 and 2050, and where each one actually stands. Read the fine print: most "net zero 2050" pledges cover Scope 1 and 2 — the company's own operations — not the emissions from burning what they sell, which is 80–90% of the total.
2030、2040、2050 年的承诺，以及各家真实的处境。请读小字：多数"2050 净零"只覆盖范围一与范围二（自身运营），不含其所售产品燃烧产生的排放——而后者占总量的 80–90%。
"Relatively ahead" means ahead of peers — not aligned with 1.5°C. Investor benchmarks find no assessed producer fully aligned.
"相对领先"只是相对同行，并不等于与 1.5°C 对齐。投资者基准评估显示，没有一家被评估的生产商完全对齐。
Beyond the producers · the rest of the chain
生产商之外 · 产业链其余环节
JPMorgan, Citi, Bank of America, Mizuho, MUFG lead fossil financing league tables year after year.
摩根大通、花旗、美国银行、瑞穗、三菱日联，年复一年位居化石燃料融资榜首。
Lloyd's syndicates, AIG, Chubb. Withdrawal of underwriting is the fastest chokepoint on new fields.
劳合社成员、AIG、丘博。撤出承保是阻断新油田最快的关口。
Export credit agencies and sovereign funds still underwrite projects the private market prices as risky.
出口信贷机构与主权基金仍在为私人市场视为高风险的项目兜底。
Accountability is no longer a matter of trust. Methane plumes are measured from orbit within days; capital plans are benchmarked line by line; courts have begun to treat corporate climate duty as law. The gap is enforcement, not evidence.
追责已经不再依赖信任。甲烷羽流可在数日内由卫星测得，资本计划被逐条对标，法院也开始把企业气候义务当作法律。缺的是执行，而不是证据。
Climate Action 100+ groups 600+ investors with tens of trillions under management. Its benchmark finds no assessed oil & gas company with capital plans aligned to 1.5°C.
Climate Action 100+ 聚集 600 多家投资者、管理规模达数十万亿美元。其基准评估显示，没有一家被评估的油气公司资本计划与 1.5°C 对齐。
Lever: votes, escalation, board challenges
杠杆：投票、升级施压、董事会挑战
Carbon Tracker on stranded assets, the TPI Centre on alignment, Global Energy Monitor on infrastructure, Urgewald's Global Oil & Gas Exit List covering 1,700+ companies, InfluenceMap on lobbying.
Carbon Tracker 追踪搁浅资产，TPI 中心评估对齐度，Global Energy Monitor 盯基础设施，Urgewald 的全球油气退出名单覆盖 1,700 多家公司，InfluenceMap 追踪游说。
Lever: naming the numbers companies omit
杠杆：把公司省略的数字说出来
UNEP's IMEO and its Methane Alert and Response System, Carbon Mapper with Tanager-1, GHGSat, and EDF's MethaneSAT which flew from 2024 to 2025. Denial has a shrinking half-life.
联合国环境署 IMEO 及其甲烷预警响应系统、Carbon Mapper 与 Tanager-1、GHGSat，以及 2024–2025 年在轨的 MethaneSAT。否认的保鲜期越来越短。
Lever: attributed, timestamped leak alerts
杠杆：可归因、带时间戳的泄漏警报
The EU Methane Regulation reaches imported gas later this decade — the first time an importer prices a producer's methane. Add CSRD, ISSB S2 and California's SB 253/261.
欧盟甲烷法规将在本十年后期覆盖进口天然气——进口方首次为生产方的甲烷定价。再加上 CSRD、ISSB S2 与加州 SB 253/261。
Lever: market access conditional on data
杠杆：以数据为前提的市场准入
More than 2,600 climate cases filed worldwide. Milieudefensie v Shell won in 2021 and was overturned on appeal in 2024: the duty of care survived, the specific 45% order did not.
全球已提起 2,600 多起气候诉讼。荷兰环保组织诉壳牌案 2021 年胜诉、2024 年上诉被撤销：注意义务被保留，45% 的具体减排命令被推翻。
Lever: precedent, disclosure, discovery
杠杆：判例、披露、证据调取
Banking on Climate Chaos, Reclaim Finance, Insure Our Future, pension-fund campaigns, and host communities from the Niger Delta to the Ecuadorian Amazon carrying the cost first.
Banking on Climate Chaos、Reclaim Finance、Insure Our Future、养老金运动，以及从尼日尔三角洲到厄瓜多尔亚马逊——最先承担代价的当地社区。
Lever: where you bank, insure and vote
杠杆：你的存款、保单与选票投向哪里
Emissions follow capital with a lag of a decade. So the fastest climate policy is often a financial one: change the price, the paperwork and the insurance, and the drilling schedule changes by itself.
排放跟随资本，滞后约十年。所以最快的气候政策往往是金融政策：改变价格、文件与保单，钻井计划自己就会变。
Yes — but only in one direction at a time, and it is currently pointed both ways. Data centres used about 415 TWh in 2024, roughly 1.5% of world electricity, and could approach 945 TWh by 2030; much of that new load is being met with gas. Meanwhile the same techniques find methane leaks in days instead of years.
能——但每次只能指向一个方向，而它现在同时指着两边。2024 年数据中心用电约 4,150 亿千瓦时，约占全球电力 1.5%，到 2030 年可能接近 9,450 亿千瓦时，其中相当部分由天然气供应。与此同时，同样的技术把甲烷泄漏的发现周期从数年缩短到数天。
IEA: widespread AI adoption could avoid on the order of 1.4 Gt CO₂e a year by 2035 — larger than data centre emissions, but not automatic.
IEA：AI 广泛应用到 2035 年每年或可避免约 14 亿吨二氧化碳当量——大于数据中心自身排放，但绝非自动实现。
AI is a multiplier, not a direction. It shrinks emissions only where regulation, procurement and disclosure point it that way.
AI 是乘数，不是方向。只有当监管、采购与披露把它指向减排，它才会减排。
Not a slogan, and not a spreadsheet either. Six things that would let anyone tell a transition from a delay — in the order they would have to happen.
不是口号，也不只是一张表。以下六件事，按它们必须发生的顺序排列，能让任何人分辨"转型"与"拖延"。