夜雨聆风
AI不是软件,而是工业革命:英伟达揭示的五层智能世界
英伟达也下场科普了，3月10日英伟达在X（twitter）上发了第一条长文，标题为“AI是一块五层蛋糕”，将AI的构成和价值进行了全方位解读，里面不乏很有价值的观点，希望在这篇文章中和大家分享。全文分为两个部分：1.英伟达“AI是块五层蛋糕”全文翻译，给了大家中英双语版，免去了大家去找英文原版的麻烦；2.不想看全文的也可以看我第二部分的解读，完全够用。下面开始：
AI Is a Five-Layer CakeAI 是一块“五层蛋糕”
AI is one of the most powerful forces shaping the world today.AI 是当今塑造世界最强大的力量之一。
It is not a clever app or a single model; it is essential infrastructure, like electricity and the internet.它并不是一个聪明的应用程序或单一模型，而是一种基础设施，就像电力和互联网一样。
AI runs on real hardware, real energy, and real economics.AI 运行在真实的硬件、真实的能源以及真实的经济体系之上。
It takes raw materials and converts them into intelligence at scale.它将原始资源转化为规模化的智能。
Every company will use it. Every country will build it.每一家公司都会使用它，每一个国家都会建设它。
To understand why AI is unfolding this way, it helps to reason from first principles and look at what has fundamentally changed in computing.要理解为什么AI 会以这种方式发展，我们需要从第一性原理出发，看看计算领域到底发生了什么根本变化。
From Pre-Recorded Software to Real-Time Intelligence
从预先编写的软件到实时智能
For most of computing history, software was pre-recorded.在计算机历史的大部分时间里，软件都是预先编写好的。
Humans described an algorithm. Computers executed it.人类描述算法，计算机执行它。
Data had to be carefully structured, stored into tables, and retrieved through precise queries.数据必须被严格结构化，存储到表格中，并通过精确查询进行检索。
SQL became indispensable because it made that world workable.SQL 变得不可或缺，因为它使这种模式的世界能够运作。
AI breaks that model.AI 打破了这种模式。
For the first time, we have a computer that can understand unstructured information.人类第一次拥有了一种可以理解非结构化信息的计算机。
It can see images, read text, hear sound, and understand meaning.它可以看图像、读文字、听声音，并理解意义。
It can reason about context and intent.它能够理解上下文和意图。
Most importantly, it generates intelligence in real time.最重要的是，它能够实时生成智能。
Every response is newly created.每一次回答都是新生成的。
Every answer depends on the context you provide.每一个答案都依赖于你提供的上下文。
This is not software retrieving stored instructions.这不再是软件在检索预先存储的指令。
This is software reasoning and generating intelligence on demand.而是软件在按需推理并生成智能。
Because intelligence is produced in real time, the entire computing stack beneath it had to be reinvented.因为智能是实时生成的，所以其底层的整个计算技术栈都必须被重新发明。
Ai as infrastructure
AI即基础设施
When you look at AI industrially, it resolves into a five-layer stack.
从产业角度看，AI可以分解为五层技术栈。
Energy
能源
At the foundation is energy.最底层是能源。
Intelligence generated in real time requires power generated in real time.实时生成的智能需要实时生成的电力。
Every token produced is the result of electrons moving, heat being managed, and energy being converted into computation.每一个token 的产生都意味着电子的流动、热量的管理以及能源向计算的转化。
There is no abstraction layer beneath this.
这之下没有抽象层
Energy is the first principle of AI infrastructure and the binding constraint on how much intelligence the system can produce.
能源是AI基础设施的第一原则，也是制约系统能产生多少智能的根本约束。
Chips
芯片
Above energy are the chips.能源之上是芯片。
These are processors designed to transform energy into computation efficiently at massive scale.这些处理器被设计为高效地将能源转化为大规模计算。
AI workloads require enormous parallelism, high-bandwidth memory, and fast interconnects.
AI工作负载需要巨大的并行性、高带宽内存和快速互连。
at the chip layer determines how fast AI can scale and how affordable intelligence becomes.
芯片层的这些特性决定了AI能以多快的速度扩展，以及智能的普及成本有多高。
Infrastructure
基础设施
Above chips is infrastructure.芯片之上是基础设施。
This includes land, power delivery, cooling, construction, networking, and the systems that orchestrate tens of thousands of processors into one machine.
这包括土地、电力供应、冷却系统、建筑施工、网络设施、以及将数万个处理器协调成一台机器的系统。
These systems are AI factories.
这些系统是AI工厂。
They are not designed to store information.
它们并非为存储信息而设计。
They are designed to manufacture intelligence.
而是为制造智能而生。
Models
模型
Above infrastructure are the models.基础设施之上是模型。
AI models understand many kinds of information: language, biology, chemistry, physics, finance, medicine, and the physical world itself.
AI模型能够理解多种类型的信息：语言、生物学、化学、物理、金融、医学，以及物理世界本身。
Language models are only one category.
语言模型只是其中的一个类别。
Some of the most transformative work is happening in protein AI, chemical AI, physical simulation, robotics, and autonomous systems.
一些最具变革性的进展正在蛋白质AI、化学AI、物理仿真、机器人技术和自主系统中发生。
Applications
应用
At the top are applications, where economic value is created.最顶层是应用，这里才产生真正的经济价值。
Drug discovery platforms. Industrial robotics. Legal copilots. Self-driving cars.
药物研发平台、工业机器人、法律辅助助手、自动驾驶汽车。
A self-driving car is an AI application embodied in a machine.
自动驾驶汽车是体现在机器中的AI应用。
A humanoid robot is an AI application embodied in a body. Same stack. Different outcomes.
人形机器人是体现在身体中的AI应用。相同的底层架构，不同的最终成果。
That is the five-layer cake:
Energy → chips → infrastructure → models → applications.
这就是五层蛋糕模型：
能源→芯片→基础设施→模型→应用。
Every successful application pulls on every layer beneath it, all the way down to the power plant that keeps it alive.
每一个成功的应用都会拉动其下方所有层级，一直延伸到维持其运行的发电厂。
We have only just begun this buildout. We are a few hundred billion dollars into it. Trillions of dollars of infrastructure still need to be built.
我们才刚刚开始这一建设进程。目前我们已投入数千亿美元，但仍有数万亿美元的基础设施有待建设。
Around the world, we are seeing chip factories, computer assembly plants, and AI factories being constructed at unprecedented scale.
在全球范围内，我们看到芯片工厂、计算机组装厂和AI工厂正以前所未有的规模建设中。
This is becoming the largest infrastructure buildout in human history.
这正在成为人类历史上规模最大的基础设施建设工程。
The labor required to support this buildout is enormous.
支持这一建设所需的劳动力非常巨大。
AI factories need electricians, plumbers, pipefitters, steelworkers, network technicians, installers, and operators.
AI工厂需要电工、管道工、管工、钢铁工人、网络技术人员、安装人员和操作员。
These are skilled, well-paid jobs, and they are in short supply.
这些都是技术熟练、薪酬优厚的工作岗位，而且目前供不应求。
You do not need a PhD in computer science to participate in this transformation.
你不需要会拥有计算机科学博士学位就能够参与这场变革。
At the same time, AI is driving productivity across the knowledge economy.
与此同时，AI正在推动整个知识经济的生产力提升。
Consider radiology.AI now assists with reading scans, but demand for radiologists continues to grow. That is not a paradox.
以放射科为例：AI现在可以协助读取扫描图像，但放射科医生的需求仍在持续增长——这并不矛盾。
A radiologist’s purpose is to care for patients. Reading scans is one task along the way.
放射科医生的职责是照顾患者，读取扫描图像只是其中一项任务。
When AI takes on more of the routine work, radiologists can focus on judgment, communication, and care.
当AI承担更多常规工作时，放射科医生可以专注于判断、沟通和护理。
Hospitals become more productive. They serve more patients. They hire more people.
医院因此变得更高效，能服务更多患者，也会雇佣更多人。
Productivity creates capacity. Capacity creates growth.
生产力创造产能，产能创造增长。
What Changed in the Last Year?
过去一年发生了什么变化？
In the past year, AI crossed an important threshold.在过去一年里，人工智能跨越了一个重要的临界点。
Models became good enough to be useful at scale.模型已经足够好，可以在大规模场景中发挥实际作用。
Reasoning improved. Hallucinations dropped. Grounding improved dramatically.推理能力得到了提升。幻觉（模型生成错误或虚假信息的情况）明显减少。模型的“事实对齐能力”（grounding，即基于真实数据和上下文生成内容的能力）得到了显著提升。
For the first time, applications built on AI began generating real economic value.人工智能应用第一次开始真正创造现实的经济价值。
Applications in drug discovery, logistics, customer service, software development, and manufacturing are already showing strong product-market fit.在药物研发、物流、客户服务、软件开发和制造业等领域，基于AI的应用已经显示出明显的产品市场匹配（Product-Market Fit）。
These applications pull hard on every layer beneath them.这些应用正在强烈拉动其底层所有技术层级的发展。
Open-source models play a critical role here.开源模型在其中发挥着关键作用。
Most of the world’s models are free.世界上大多数模型都是免费的。
Researchers, startups, enterprises, and entire nations rely on open models to participate in advanced AI.研究人员、初创公司、企业甚至整个国家，都依赖开源模型参与到先进人工智能的发展之中。
When open models reach the frontier, they do not just change software.当开源模型达到技术前沿时，它们改变的不仅仅是软件。
They activate demand across the entire stack.它们会激活整个技术栈的需求。
DeepSeek-R1 was a powerful example of this.DeepSeek-R1就是一个强有力的例子。
By making a strong reasoning model widely available, it accelerated adoption at the application layer and increased demand for training, infrastructure, chips, and energy beneath it.通过让一个强大的推理模型广泛可用，它加速了应用层的采用，同时也增加了对训练、基础设施、芯片和能源等底层资源的需求。
What This Means
这意味着什么
When you see AI as essential infrastructure, the implications become clear.当你把人工智能视为一种关键基础设施时，其影响就会变得清晰。
AI starts with a transformer LLM. But it’s much more.人工智能始于Transformer架构的大语言模型。但它远不止如此。
It is an industrial transformation that reshapes how energy is produced and consumed, how factories are built, how work is organized, and how economies grow.它是一场工业级转型，将重新塑造能源的生产与消费方式、工厂的建设方式、工作的组织方式以及经济增长的方式。
AI factories are being built because intelligence is now generated in real time.AI工厂正在被建设，因为“智能”现在可以被实时生产。
Chips are being redesigned because efficiency determines how fast intelligence can scale.芯片正在被重新设计，因为效率决定了智能能够以多快的速度扩展。
Energy becomes central because it sets the ceiling on how much intelligence can be produced at all.能源变得至关重要，因为它决定了智能生产的上限。
Applications accelerate because the models beneath them have crossed a threshold where they are finally useful at scale.应用之所以迅速发展，是因为其底层模型已经跨过了一个临界点——它们终于能够在大规模环境下发挥作用。
Every layer reinforces the others. This is why the buildout is so large.This is why it touches so many industries at once.每一个层级都会强化其他层级。这就是为什么当前的建设规模如此庞大。这也是为什么它会同时影响如此多的行业。
And this is why it will not be confined to a single country or a single sector.这也解释了为什么它不会局限于某一个国家或某一个行业。
Every company will use AI. Every nation will build it.每一家公司都会使用人工智能。每一个国家都会建设它。
We are still early. Much of the infrastructure does not yet exist.我们仍然处于早期阶段。大量基础设施尚未建成。
Much of the workforce has not yet been trained. Much of the opportunity has not yet been realized.大量劳动力还没有接受相关训练。大量机会仍然没有被实现。
But the direction is clear. AI is becoming the foundational infrastructure of the modern world.但方向已经非常清晰。人工智能正在成为现代世界的基础设施。
And the choices we make now, how fast we build, how broadly we participate, and how responsibly we deploy it, will shape what this era becomes.而我们现在所做的选择——建设速度有多快、参与范围有多广、部署方式有多负责任——将决定这个时代最终会变成什么样。
以上就是英伟达的x长文的全部内容，全文最核心的以及想要表达的是未来世界一定是AI作为基础设施的世界，不布局AI就是在丢掉未来，谁掌握AI基础设施，谁掌握未来经济。当然，英伟达一定会有这样的主张，不然芯片卖给谁？当然这并不代表他说的就仅仅具有营销的价值。另外，他还强调了AI工业对人类劳动力的需求，这实际上是在消除社会的恐慌和前段时间美国国内民众对AI的抵制，这部分宣传的意义更大。总的来说，我认为全文最有价值的是：
AI其实是“能源产业”
AI的本质可以写成公式：AI = Energy →Computation →Intelligence
换句话说：AI本质上是把电变成智能的机器，这意味着：未来 AI 竞争的核心不是模型，而是电力、芯片、数据中心这类物理基础设施。
这也是为什么：美国疯狂建数据中心、中东投电力 + AI、中国建算力中心的原因。
AI是一场“工业革命”，不是软件革命
文章最核心的一句话其实是：
AI factories manufacture intelligence
AI工厂制造的是：智能，这与传统软件不同。
传统软件：程序员→写代码→软件
AI时代：能源→GPU →数据中心→模型→智能
这更像钢铁工业，而不是软件工业
未来最大的瓶颈是能源
现在全球AI最大的限制其实是：电
例如：一个大型AI数据中心，功耗：500MW≈一个中型城市，未来 AI 规模扩大 10 倍，意味着能源体系必须升级。
所以未来会出现三个超级产业：核电、电网、AI算力。
AI的五层结构其实少了一层
我认为AI其实是六层结构：
Energy→Chips→Infrastructure→Models→Applications→Society，最后一层是：社会结构。
AI将改变：整个社会基本运行逻辑，从结构开始：劳动力结构、教育结构、知识信用体系、金融结构、社会财富分配等等。
往期回顾：
