2024 was the first year energy-transition investment crossed two trillion dollars. That is a real achievement, and it deserves to be said plainly: humans built the fastest-scaling industrial curve in modern history — solar, batteries, electric vehicles — mostly on private capital. 2024 年是能源转型投资首次突破两万亿美元的一年。这是真实的成就，应当直说：人类用私人资本推动了现代史上扩张最快的一条产业曲线——光伏、电池、电动车。
But look at where the money lands. Roughly seven of every ten dollars go to two mature, bankable places. The layers that decide whether 2035 actually works — grids, storage, industrial heat, water, cooling, adaptation — take the change left in the till. 但看资金的落点：每十美元里约有七美元流入两个已经成熟、可融资的领域。真正决定 2035 年能否运转的那些层——电网、储能、工业热、水、制冷、适应——只拿到抽屉里剩下的零钱。
This station reads four horizons against the published data, marks what is on track and what is not, and names the things that do not exist yet but must be paid for now. 本站以公开数据读取四个时间区间，标注哪些达标、哪些偏离，并指出那些今天还不存在、却必须现在就付钱去建的东西。
“Readiness is not a date. It is a purchase order signed nine years early.” 「所谓准备好，不是一个日期，而是一张提前九年签下的采购订单。」
Share of the ~US$2.08 trillion of energy-transition investment recorded in 2024, by sector. Bars are scaled to the largest category.2024 年约 2.08 万亿美元能源转型投资的分部门占比。条形以最大类目为基准缩放。
Read it this way: the two categories with proven consumer or utility revenue take ~71%. The four categories that unblock everything else — storage, clean industry, hydrogen, capture — together take about 4.2%. The transition is being financed at its easiest edge, not at its bottleneck. 这样读：拥有成熟消费端或电力端收入的两类占约 71%；而真正为其他一切解锁瓶颈的四类——储能、清洁工业、氢、碳捕集——合计仅约 4.2%。转型的融资集中在最容易的一端，而不是卡口。
Not the totals. The ratios. Each one is a design flaw in how capital is allocated, not a shortage of capital.重要的不是总量，而是比率。每一个都是资本配置的设计缺陷，而非资本短缺。
For every dollar into fossil energy supply, only about the same again goes into low-carbon supply. BNEF's own alignment marker is 4:1 by 2030. We are at roughly a quarter of the required ratio.每一美元投向化石能源供给，仅有约等量资金投向低碳供给。BNEF 的对齐标准是 2030 年达到 4:1。我们只有所需比率的约四分之一。
Developing countries need US$215–387bn a year for adaptation by 2030. International public adaptation finance reached about US$27.5bn in 2022 — a gap of US$187–359bn every year. Adaptation is roughly 5% of all tracked climate finance.发展中国家到 2030 年每年需要 2150–3870 亿美元用于适应；2022 年国际公共适应资金约 275 亿美元，年缺口 1870–3590 亿美元。适应约占全部可追踪气候资金的 5%。
Emerging and developing economies outside China hold around two-thirds of the world's people and receive roughly 15% of global clean energy investment. Capital costs there run two to three times higher.除中国外的新兴与发展中经济体承载着全球约三分之二人口，却只获得约 15% 的清洁能源投资，且当地资本成本高出 二到三倍。
Same question in each period: is capital arriving where the physics and the risk are?每个区间都问同一个问题：资本是否到达了物理规律与风险所在之处？
Annual transition investment roughly doubled, from about US$1tn to US$2.08tn. Solar and battery costs fell faster than any official forecast. And yet: early-stage climate-tech venture funding fell around 40% from its 2021 peak just as deployment accelerated, adaptation stayed a rounding error, and emissions kept rising. Capability was demonstrated; sequencing was not. 年度转型投资大致翻倍，从约 1 万亿美元增至 2.08 万亿美元；光伏与电池成本下降速度超过所有官方预测。但与此同时，早期气候科技风险投资相比 2021 年高点下降约 40%——恰好发生在部署加速之时；适应投入仍是零头；排放继续上升。能力得到了验证，次序没有。
Grid investment sits near US$400bn a year and needs to pass US$600bn a year by 2030 simply to absorb what is already being built. Thousands of gigawatts of clean projects wait in interconnection queues. Meanwhile AI and data-centre electricity demand is set to more than double to about 945 TWh by 2030 — new load arriving faster than new wires. This is the decade where a good decade of generation becomes a wasted one. 电网投资约为每年 4000 亿美元，到 2030 年需超过每年 6000 亿美元，才能仅仅吸纳已经在建的电源。数千吉瓦清洁项目排队等待并网。同时，人工智能与数据中心用电到 2030 年将翻倍以上，约达 945 太瓦时——新负荷来得比新线路更快。这十年决定了一个好的电源十年会不会被浪费。
Tracked climate finance reached about US$1.46tn in 2022; the requirement rises to roughly US$10tn a year from 2031. The industries that must carry this decade currently receive under one percent of transition capital combined. These are 20–40 year assets with one or two investment windows in a lifetime — and the window for plants running in 2040 opens before 2030. 2022 年可追踪气候资金约 1.46 万亿美元；而 2031 年起的需求升至每年约 10 万亿美元。必须承担这十年的重工业目前合计获得不足 1% 的转型资本。它们是寿命 20–40 年的资产，一生只有一两次投资窗口——而 2040 年运转的工厂，其窗口在 2030 年前就要打开。
In 2024 natural catastrophes caused roughly US$318bn of economic losses, of which about US$137bn was insured — well over half uninsured. That protection gap is the market's clearest, least-acted-upon price signal: unfunded resilience does not disappear, it re-appears as claims, relocation, crop failure and downgrades. Meanwhile fossil fuel subsidies of about US$7tn a year (IMF, explicit and implicit) remain the largest pool of transition capital nobody has moved. 2024 年自然灾害造成约 3180 亿美元经济损失，其中约 1370 亿美元有保险覆盖——超过一半没有保障。这个保障缺口是市场上最清晰、却最少被行动回应的价格信号：没有被投资的韧性不会消失，它会以理赔、迁移、歉收与评级下调的形式回来。同时，每年约 7 万亿美元的化石燃料补贴（IMF，含显性与隐性）仍是无人挪动的最大一笔转型资本池。
These are genuine human wins. Each is now limited by something adjacent that nobody funded.这些是真实的成绩。每一项如今都被一个相邻的、无人出资的环节所限制。
The only curve that beat every forecast for two decades running. Record additions, still falling costs. Limited now by interconnection, curtailment and the missing hours after sunset.二十年间唯一持续超越所有预测的曲线：装机屡创纪录、成本继续下降。如今受制于并网、弃电，以及日落之后缺失的那些小时。
world-class → bottlenecked downstream世界级 → 下游受阻Around one in four new cars sold globally is now electric — a consumer transition delivered largely by markets. Charging density, distribution grids and battery supply chains have not kept pace.全球每四辆新车中约有一辆是电动车——这是一次主要由市场完成的消费端转型。但充电密度、配电网与电池供应链没有同步跟上。
IEA Global EV Outlook 2025IEA《2025 全球电动车展望》Cost declines have been extraordinary. Investment has not followed: about US$54bn in 2024, some 2.6% of transition capital, for the asset class that converts variable power into firm power.成本下降幅度惊人，投资却没跟上：2024 年约 540 亿美元，占转型资本约 2.6%——而这正是把波动电力变成可靠电力的资产类别。
technology ready · capital thin技术已就绪 · 资本过薄Green bonds, blended structures, guarantees, taxonomies — the instruments exist and work. They are simply not pointed at the highest-risk, highest-need geographies at the volume required.绿色债券、混合融资结构、担保、分类目录——工具已存在且有效，只是没有以所需规模指向风险最高、需求最大的地区。
instruments exist · aim is wrong工具具备 · 瞄准偏了Global climate science and corporate reporting improved enormously. What still barely exists is decision-grade, machine-readable, asset-level resilience data an underwriter or lender can act on tomorrow morning.全球气候科学与企业报告已大幅进步。仍然几乎不存在的，是承保人或贷款人明天早上就能据以行动的、可机读的资产级韧性数据。
reporting ≠ decidable data披露 ≠ 可决策数据The cheapest tonne available, technically settled, and repeatedly stalled by policy whiplash and split incentives between landlord, tenant and utility.最便宜的一吨减排，技术早已成熟，却屡屡因政策反复以及业主、租户与电力公司之间的利益错配而停滞。
cheapest tonne · least financed最便宜的一吨 · 最少融资Ordered by how much other capital each one unlocks per dollar spent.按“每投入一美元能解锁多少其他资本”排序。
Stop treating transmission as plumbing behind the project and treat it as the primary climate asset. Interconnection reform, anticipatory build, HVDC corridors, and paying for grid capacity ahead of demand rather than after complaint.不要再把输电当成项目背后的“管道”，它就是最主要的气候资产。改革并网机制、前瞻性建设、直流输电走廊，并在需求到来之前而不是抱怨之后就为电网容量付费。
The same solar farm costs two to three times more to finance in the countries that need it most. Guarantee-first public capital, local-currency lending and FX hedging at scale move more tonnes than another subsidy for mature markets.同一座光伏电站，在最需要它的国家融资成本高出二到三倍。以担保为先的公共资本、本币贷款与规模化的汇率对冲，比再给成熟市场一笔补贴能带来更多减排。
Adaptation is stuck at roughly 5% of climate finance because it is framed as aid, not as revenue. Avoided-loss contracts, water tariffs, resilience-linked insurance discounts and cooling-as-a-service give it a cash flow.适应长期停留在气候资金的约 5%，因为它被当作援助而非收入来设计。避免损失合约、水价机制、与韧性挂钩的保费折扣、制冷即服务，都能让它产生现金流。
Demand-side certainty before supply-side subsidy: aggregated procurement pools, contracts for difference, green-premium offtake from buyers who actually build things. Under 1% of flows today decides steel and cement for 2045.先给需求端确定性，再谈供给端补贴：集合采购池、差价合约、由真正在建造实物的买方承担绿色溢价。今天不足 1% 的资金流，决定了 2045 年的钢铁与水泥。
About US$200bn a year flows to nature-based solutions while roughly US$7tn a year flows in nature-negative directions. The leverage is in the second number, in lending policy and subsidy design, not in a bigger offset market.每年约 2000 亿美元流向基于自然的解决方案，而约 7 万亿美元流向对自然有害的方向。杠杆在第二个数字上——在贷款政策与补贴设计里，而不是在一个更大的抵消市场里。
A protection gap above half of all catastrophe losses is the truest price signal we have. Parametric cover, sovereign and city-level pools, and open hazard data turn resilience from a cost line into an underwritable asset.超过一半灾害损失没有保障，这是我们拥有的最真实的价格信号。参数化保险、主权与城市层级的风险池、开放的灾害数据，能把韧性从成本项变成可承保的资产。
Around US$7tn a year of fossil support is the single largest identified source of transition finance on earth. It does not need to be raised. It needs to be moved.每年约 7 万亿美元的化石能源支持，是地球上已识别的最大一笔转型资金来源。它不需要被“筹集”，它需要被“挪动”。
Each item has a lead time longer than the time we are giving it. Waiting for 2035 to arrive is the same as choosing not to be ready.每一项的建设周期都长于我们给它的时间。等着 2035 年到来，等于选择了不准备好。
Order books for high-voltage equipment already stretch years out. No new factory, no new grid — regardless of how much project equity is available.高压设备订单已经排到数年之后。没有新工厂就没有新电网——无论项目股权有多充裕。
Long-duration storage, geothermal, hydro flexibility and nuclear uprates: the products that make a solar-heavy grid dependable in the worst week of the year.长时储能、地热、水电灵活性与核电增容：让以光伏为主的电网在全年最糟糕的一周仍然可靠的那些产品。
Machine-readable, interoperable, asset-level verification of energy, emissions and resilience that an insurer, lender and regulator can all read from the same file. Without it, capital prices uncertainty instead of risk.对能源、排放与韧性的资产级验证，必须可机读、可互操作，让保险公司、贷款人与监管者读同一份文件。没有它，资本只能给“不确定性”定价，而不是给风险定价。
Urban heat is now the deadliest climate hazard in many cities, and it has no infrastructure budget line: shade, district cooling, cold chain, grid capacity for the hottest 40 hours of the year.城市高温已成为许多城市最致命的气候风险，却没有一条基础设施预算科目：遮阴、区域供冷、冷链，以及为一年中最热的 40 个小时准备的电网容量。
Reuse, leak reduction, clean-powered desalination, irrigation retrofits and basin-level governance. Water failure arrives before energy failure in most economies.再生水、降低漏损、清洁能源驱动的海水淡化、灌溉改造与流域治理。在多数经济体，水的失效比能源的失效来得更早。
Not another demonstration unit. Whole-site electrification and heat retrofits with bankable offtake, at the scale of an existing mill or kiln line.不是又一个示范装置，而是整厂电气化与工业热改造，配以可融资的承购，规模对应现有的轧机或窑线。
Money that moves on a forecast, not on a damage assessment: parametric triggers, contingent credit lines, city and sovereign resilience facilities agreed years in advance.依据预报而不是灾损评估就能启动的资金：参数化触发、应急信贷额度，以及提前数年谈定的城市与主权韧性工具。
Grid engineers, high-voltage linespeople, commissioning crews, heat-pump installers, retrofit surveyors. Capital cannot be deployed faster than trained hands.电网工程师、高压线路工、调试团队、热泵安装工、改造勘察员。资本的落地速度不可能快于受过训练的双手。
The next generation of investable organisations in this space is small, measurement-native and shaped like infrastructure software rather than a consultancy or a fund.这一领域下一代可投资的组织：团队小、以测量为本，形态更像基础设施软件，而不是咨询公司或基金。
Six to twenty people producing the throughput of a two-hundred-person firm. Unit economics that work at small scale, and improve — not economics that require scale to survive.六到二十人做出两百人机构的产出。单位经济在小规模下就成立、并持续改善——而不是必须靠规模才能存活。
Own the interface, the standard, the schema everyone else must read and write. Interoperability is the moat; lock-in is the tell of a weaker business.拥有接口、标准与数据模式，让其他人必须读写它。互操作性才是护城河；锁定客户往往是较弱生意的标志。
Verification is the product, not the annual report. Every claim machine-checkable, timestamped, and good enough for an underwriter — because the missing data layer is the actual market.验证本身就是产品，而不是年报的一章。每一项声明都可机器核验、带时间戳，并足以支撑承保——因为缺失的数据层才是真正的市场。
Hardware and project assets held in separate vehicles; the company itself stays a high-margin core. Investors want physical impact without buying a balance sheet they cannot underwrite.硬件与项目资产放在独立载体中，公司本体保持高毛利内核。投资者要的是物理世界的实效，而不是一张无法承保的资产负债表。
Utility, grid operator, insurer, port, city, industrial offtaker. Procurement-ready from day one: security review, data residency, audit trail, contract terms that survive a legal department.电力公司、电网调度、保险公司、港口、城市、工业承购方。第一天就具备可采购性：安全审查、数据驻留、审计轨迹，以及能通过法务部门的合同条款。
Dollars per tonne, per megawatt, per kilowatt-hour, per cubic metre, per avoided claim — stated before the narrative. Thermodynamics does not respond to a pitch deck.每吨、每兆瓦、每千瓦时、每立方米、每避免一笔理赔的成本——在叙事之前先讲清楚。热力学不会回应一份融资演示。
Permitting, interconnection rules, tariff structures and disclosure regimes understood in-house. In this sector the regulatory path is the product roadmap.许可流程、并网规则、电价结构与披露制度，团队内部就懂。在这个行业，监管路径就是产品路线图。
AI aimed at grid studies, permit packages, claims adjudication, retrofit surveys, portfolio hazard screening — the slow, expensive, human-bottlenecked steps that currently delay deployment by years.把 AI 用在电网研究、许可文件、理赔判定、改造勘察、组合风险筛查上——正是这些缓慢、昂贵、卡在人力上的步骤，让部署延后了数年。
Every installation, inspection and claim makes the model measurably better, and the improvement is legible to a buyer. That is the compounding asset — not headcount.每一次安装、巡检与理赔都让模型可度量地变好，而这种进步能被买方看懂。这才是复利资产——而不是人数。
Verdict, honestly held: the direction is good, the volume is improving, and the allocation is wrong in a way that compounds. Capital is behaving rationally against the risk it can see and measure — which is exactly why the missing measurement layer, the missing wires and the missing resilience revenue models are the highest-return places to intervene. 诚实的判断：方向是对的，总量在改善，但配置的偏差会不断累积。资本对它能看见、能度量的风险作出了理性反应——正因如此，缺失的测量层、缺失的线路、缺失的韧性收入模式，才是回报最高的介入点。
2035 is not a deadline to wait for. It is a construction schedule that has already started without us. 2035 不是一个等待中的截止日，而是一份已经在没有我们的情况下开始的施工计划。
Take the 2035 build list to your investment committee把 2035 建设清单带进你的投委会Figures are as published; totals are approximate and rounded. Where two sources use different scopes, both are named.数字以原始出版为准，总量为近似并已取整。当两个来源统计口径不同时，两者均已列出。