Author: nini

If you missed out on the Apple supply chain in 2010, missed out on the Tesla supply chain in 2020, and even regretted missing the Nvidia supply chain in the past two years—

the SpaceX supply chain is just getting started.

Of course, I think it doesn’t seem worthwhile to simply chase after SpaceX itself. Its stock rose 19% on the first day of its IPO, priced at 135 and soaring to 160, with a sales price ratio close to 100 times, and the company is still in huge losses. Retail investors rushing in on the IPO day face significant pressure.

So what I want to talk about is the companies that supply it.

History has repeatedly verified the same logic: super terminals crazily replenish the underlying supply chain. In 2010, when Apple released the iPhone 4, Luxshare Precision had a revenue of 1 billion, and ten years later, it soared to 92.5 billion, with a stock price increase of 30 times. In 2019, when Tesla’s Shanghai factory began production, CATL had a market value just over 100 billion, and five years later, it surpassed 1 trillion. Nvidia has skyrocketed these past two years, and Zhongji Xuchuang’s market value increased from tens of billions to over a thousand billion.

Apple, Tesla, Nvidia—each time the super terminal is front and center, but what truly allows a group of people to earn big money are the supply chain companies behind them.

SpaceX spends hundreds of billions each year on purchasing chips, materials, parts, and industrial gases. These purchasing orders gradually become real revenue for certain companies. After the prospectus was made public, this supply chain finally has traceable data.

Let’s first look at where the money comes from and where it goes for SpaceX

Its business mainly consists of these three parts. The first part, Starlink. Last year’s revenue was 11.3 billion USD, accounting for 60% of the group, with over 10 million global subscribers—this is the only stable profit-making part of SpaceX, which can even be said that all the cash-burning areas rely on it to cover the costs.

The second part, rockets. Falcon and Starship have an annual R&D investment of 3 billion USD, resulting in the lowest cost of commercial launches globally, with plans for 100 launches in 2026 and a demand for 1,500 Raptor engines. The third part, AI. Last year it lost over 6 billion, and it is currently building the Colossus supercomputer, stacking 220,000 GPUs, with plans for orbital data centers in the sky.

So, the flow of money is quite simple: money earned from Starlink → invested in rockets to reduce launch costs → low-cost launches send AI hardware into space → AI computing power is rented out to generate income. It’s roughly a cycle like this.

This cycle releases hundreds of billions of procurement orders annually, so where is this money going?

Divided by whether they can be replaced, suppliers fall into three categories.

First Category: Irreplaceable in the short term

1. Nvidia, all 220,000 GPUs for the Colossus supercomputer are theirs. But Nvidia's real moat isn’t hardware; it’s CUDA, the software ecosystem that virtually all AI training uses to write code globally. Hardware can be switched, but the migration cost for a decade’s worth of code is not something that can be compensated in just one or two years. We can understand that as long as SpaceX is still building supercomputing, Nvidia will be getting paid.
2. Eutelsat, code SATS. It holds the radio frequency spectrum for satellite communications. What is the spectrum? You can think of it as lanes in the sky, governed by physical laws to the few that exist; whoever occupies it first owns it, and no matter how strong your technology is, you cannot create it out of thin air. Musk's phone-to-satellite connection feature must pass through it. Without paying the toll, the signal will collide with signals from other satellites. Moreover, SATS holds about 3% of SpaceX shares. The day before the IPO, it rose 11%, with options trading volume at 11 times its usual amount.
3. Filtronic, code FTC, listed in London, note that you can’t find it in U.S. stock markets. It provides millimeter-wave signal amplifiers for Starlink satellites, enabling clearer and farther communication. In 2024, it signed a 47.3 million GBP contract, with SpaceX contributing 83% of its revenue, and it received up to 10% subscription rights. This may seem trivial, but aerospace-level certification requires years of repeated testing under vacuum, radiation, and extreme temperature changes. Once certified, SpaceX won't easily switch suppliers because the recertification period cannot keep up with production acceleration. Also, Filtronic’s stock price nearly doubled in a year.
4. Materion, code MTRN. The only integrated producer of beryllium from ore to finished product globally, controlling about 56% of the supply. Beryllium is one-third lighter than aluminum, six times stronger than steel, with a melting point of nearly 1300 degrees Celsius—few metals on Earth meet all three conditions of being light, hard, and heat-resistant. The F-35 fighter jet, the Webb Space Telescope’s lenses, and Starship’s structural components all use it. The U.S. Department of Defense has listed beryllium as a strategic material, and Materion has been the exclusive certified supplier for the F-35 for over a decade, demonstrating its scarcity.
5. STMicroelectronics, code STM. It manufactures phased array antenna chips for SpaceX, having delivered over 5 billion units, covering more than 10,000 satellites. STM predicts that the low-earth satellite business can reach 2 billion USD by 2028 and 2.9 billion USD by 2030.

Second Category: Technically replaceable but the cost of switching is too high

1. Honeywell, code HON. It controls the flight control and inertial navigation system of rockets—what tells the rocket where it is, where it’s flying, and how to maintain its posture is entirely controlled by it. This certification has been built up over decades from Apollo to the space shuttle to commercial spaceflight. Switching suppliers would mean having to re-implant the rocket’s brain, rewriting all the underlying code, and starting the new certification process from scratch. SpaceX launches over a hundred times a year; it cannot afford to halt launch schedules just to save on procurement costs.
2. Carpenter Technology, code CRS. It refines special steel alloys for the Raptor engines. Vacuum melting and repeated purifying, controlling impurities to millionths of a level. Just the slightest difference can lead to disaster in the combustion chamber. This material and process cannot be transmitted based solely on drawings; constructing a comparable production line could take decades or more.
3. Hexcel, code HXL. It supplies aerospace carbon fiber; every kilogram less on the rocket means one kilogram less of effective payload. Carbon fiber skeletons are half as light as metals, maintaining strength. They have cooperated with SpaceX for over a decade, and the material formula and weaving process are specifically tailored to SpaceX’s needs. If they switch manufacturers, the entire material system must be validated from scratch.
4. Broadcom AVGO, responsible for the 10Gbps data exchange between earth and space. High-speed data diversion without congestion relies on it. Linde Group invested 100 million USD near the SpaceX Starbase in Texas to build an air separation plant, exclusively providing liquid oxygen and liquid nitrogen due to the large amount of high-purity industrial gases consumed during rocket launches; the closer the distance, the lower the cost. This site selection itself acts as a moat.

Third Category: Requires stable mass production and cost minimization

You may not have seen the physical Starlink dish, but think about it, they need to deploy a total of 30 million units globally. Each unit contains thousands of components and undergoes dozens of processes, needing to be produced on assembly lines like smartphones while also withstanding aerospace-grade vibrations and temperature variations.

At this scale, technical factors become less critical; the most important thing is who can stably deliver and who can minimize costs.

The logic of Foxconn working for Apple applies identically here. QIQI Technology, code 6285, is the largest OEM for Starlink terminals and routers globally. The quality control standards were developed through years of collaboration with SpaceX, so it is not just any factory that can take over.

There are several A-share companies that qualify. Xinwei Communication, code 300136, is the sole global supplier of high-frequency connectors for Starlink terminals, with SpaceX-related orders projected at about 1.05 billion USD by 2025. Paike New Materials, code 605123, is the only Chinese supplier of forged components for Starship body and engines, with orders around 680 million, making up 35% of the company's revenue. Western Materials, code 002149, is the exclusive supplier of niobium alloys for Raptor engines, with orders of about 1.02 billion. Yingliu Holdings, code 603308, supplies core castings for Raptor turbo pumps, accounting for 42% of its revenue—SpaceX orders have already become this company’s largest income source.

Going smaller, Tianyin Electromechanical can be likened to the star sensor on Starlink satellites; satellites rely on it to observe stars to determine their posture, with a market share exceeding 60%. Tongyu Communication makes the ground antenna modules for Starlink, expecting orders of 300 million by 2026.

On the U.S. side, there are also a few companies. Trimble, code TRMB, manages timing; thousands of satellites flying above must have their clocks synchronized to the same beat—missing even a microsecond can lead to communication errors. Astronics, code ATRO, handles rocket power distribution. CTS, code CTSH, manages heat dissipation. None of these are black technologies, but they are all essential screws in the overall system.

You might ask, why now for these companies that have always been here?

Three reasons.

• First, the procurement volume has just started to increase. Plans for 100 launches in 2026, with Starship accelerating testing, and AI data centers starting deployment into space in 2028. The target for Starlink terminals is 30 million, and there are currently only 10 million subscribers. The pace at which SpaceX is spreading money has not yet peaked.
• Second, transparency has opened up for the first time. Previously, SpaceX was a private company; procurement data was a black box. After the prospectus was made public, quarterly and annual reports will continue to disclose data, and the order growth rate for supply chain companies can be tracked and verified.
• Third, looking back at historical rhythms. The Apple supply chain took ten years from iPhone 4 to its peak. The Tesla supply chain has taken seven years from Model 3 mass production to now. The current position of the SpaceX supply chain resembles Tesla in 2018, where mass production has just begun, suppliers are just being established, and the order growth rate is just starting to steepen. Meanwhile, Starship is still in testing, Starlink is gradually expanding, and the AI data center has not yet been constructed, which is currently akin to SpaceX’s 2018.

Finally

Buying SpaceX on its IPO day, I believe, is paying for Musk’s dream, and it is a very high-priced space dream. Of course, you could also say you simply believe in Musk, which is also your dream.

But perhaps we can look at it from a different angle,

in terms of the supply chain, we bet on something else, because no matter how SpaceX stock prices move, its hundreds of billions in procurement orders each year will always need to be fulfilled by someone. These orders are unrelated to stock prices; they represent consistent revenue arriving monthly.

This article does not constitute investment advice. There are still some issues, such as the cyclical nature of beryllium, geopolitical discounts for Taiwanese manufacturers, insufficient liquidity for smaller companies, and certifications potentially being reshuffled due to technological iterations. Each company needs to be judged individually.

However, if you didn’t get a share allocation on the day SpaceX went public,

you can take a different approach; instead of chasing high prices, let’s take a look at those quietly supplying.

The giant has already ignited the engines; this time, the shovel is in a place you can reach~