On March 21, Austin. Elon Musk held a press conference in an abandoned power plant, announcing something that no one in the AI industry dared to do: to make chips himself.

Tesla, SpaceX, and xAI jointly built Terafab—a $25 billion 2-nanometer chip manufacturing plant. Everything from chip design, photolithography, manufacturing, packaging, to testing will be completed under one roof. The first product, Tesla AI5, has performance close to the NVIDIA H100, with purported inference costs being 10 times cheaper, with samples planned for the end of 2026 and mass production in 2027. The target capacity: 1 petawatt of AI computing power per year—50 times the current total AI computing power globally.

80% of the chips will be sent to space, operating AI on satellites and distributed back to Earth via Starlink. Only 20% will remain on Earth. The power required for 1 petawatt of computing cannot be supplied by the Earth’s power grid—solar radiation in space is 5 times that on the ground, and the vacuum has higher heat dissipation efficiency, without grid bottlenecks.

As soon as the news broke, the industry erupted.

He is challenging a giant that no one dares touch.

The AI chip market is essentially an absolute monopoly held by NVIDIA.

The global capital expenditure for AI infrastructure is expected to be $400-450 billion this year, with chip procurement accounting for $250-300 billion. The vast majority of this money flows to one company—NVIDIA. It has a backlog of $1 trillion in orders, a market value exceeding $3 trillion, and a gross margin of over 55%. The H100 chip sells for $30,000 and is in high demand.

The entire AI industry is being held hostage by one company; everyone knows that. Google has TPU but only uses it for itself, AMD is chasing but still far behind, and Intel's attempts at contract manufacturing have been disastrous. No one has truly stepped up and said: I'm going to fight you head-on.

Then Musk stepped up. Designing and manufacturing for himself, even building the factory.

What is he really thinking?

The surface reason is easy to understand—there are not enough chips. Tesla's autonomous driving, robotaxi, Optimus robots, and xAI's Grok are all power-hungry monsters. Queuing up to buy H100 from NVIDIA? Buying in bulk doesn't guarantee that you will get it. Looking for TSMC to make them? Apple is ahead in the queue.

But $25 billion is too heavy just to solve the supply chain issue. What he sees is much bigger than "buying chips."

He lays his cards on the table: Tesla has millions of vehicles and robots on the ground. xAI has the Grok large model. SpaceX has rockets capable of sending things into space. Starlink has a global satellite network to send data from the sky back to Earth. Now Terafab fills the last piece—chip manufacturing.

From making chips to running AI models, to launching into space, to global distribution. The entire chain is in one person's hands.

The last person to do something like this was Rockefeller—controlling the entire oil supply chain from extraction to refining to transportation to retail, becoming the most powerful person of that era. This time, the resource has shifted from oil to computing power.

Can this be done? Opinions vary.

This matter is highly controversial in the industry. Supporters argue that Musk has done many "impossible" things before—SpaceX, Starlink, Tesla, and each time he faced skepticism, he ultimately succeeded. Wedbush set a target price of $600 for Tesla.

Critics have equally strong reasons. Jensen Huang made a very subtle comment: "Building your own chip factory is an extremely difficult challenge." TSMC's processes have been accumulated over decades, with know-how spanning over 2,000 steps, which cannot simply be bought with money. Bernstein calculated a number: achieving the 1 petawatt target could end up costing $5 trillion. Moreover, every advanced process chip factory in the past decade—TSMC in Arizona, Samsung in Taylor, Intel in Ohio—has exceeded budget and faced delays. There are no exceptions.

Whether it can be achieved, no one knows at the moment. But a more worth considering question is—

If it succeeds: a disruptive landscape.

First, NVIDIA's monopoly will be pierced.

Everyone in the industry has felt the squeeze from NVIDIA on the entire AI market. If a substitute chip appears on the market with close performance and inference costs that are 10 times cheaper, even if it only serves Musk's own company, it means NVIDIA loses a super client. The competitive pressure will force NVIDIA to lower prices or accelerate innovation. The price of AI computing power may therefore decline.

Second, AI computing power will move from the ground to space.

If the plan for 80% of the chips to go to space is realized, the physical foundation for AI computing will change. Previously it was assumed that computing power resided in data centers, constrained by the power grid, heat dissipation, and land availability. Deploying in space breaks this ceiling. Thousands of AI satellites operating in orbit, selling computing power to the entire world through Starlink—this business model has enormous potential.

Third, the power structure of the AI industry will be reshuffled.

The current distribution of power in AI is: NVIDIA makes chips, TSMC manufactures them, Meta/Google/OpenAI develop models, and AWS/Azure provide cloud services. There are different players at each layer. But if Musk integrates chip manufacturing + AI models + space deployment + global distribution, he crosses all four layers. This kind of vertical integration will make all players uncomfortable.

Lastly, geopolitical considerations. Currently, over 90% of advanced chips globally are produced by TSMC, which has factories in Taiwan. If something happens in the Taiwan Strait, global AI will come to a halt. Terafab built on U.S. soil is a $25 billion investment for Washington, acquiring domestic advanced chip production capacity, and this trade-off seems worthwhile by any calculation.

My view.

Will it be completed on time? The likelihood of delays is high. Ultimately, how much will it cost? It will likely far exceed $25 billion.

But the direction is correct.

Musk is the only person globally currently holding chip manufacturing, AI models, rocket launches, global satellite communication, and millions of hardware terminals simultaneously. Each of these five pieces is a trillion-dollar business, and others lack the conditions to replicate this.

The power map of the 20th century was drawn with oil. This century is being redrawn with computing power.

Musk has circled a large area on this new map. Whether anything fruitful comes from it will depend on the next few years. But the chosen position is indeed good.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。