Technology invades the modern world

Chapter 356 Cosmic Project

The best innovation is undoubtedly the kind that stems from technological advancements that lead to leapfrog development and, consequently, innovation itself.

But this is very difficult and cannot be effective in the short term.

Different technological integrations can achieve results far superior to the original technologies. For example, Li Auto integrated refrigerators, color TVs, and large sofas into new energy vehicles. This is also an innovation that has achieved great commercial success.

What Lin Ran needs to do is a more difficult technology integration, because even though the existing technology is mature, theoretically it is feasible to turn the moon into a data center and do so for several months, but there are still countless problems to overcome.

Moreover, the current technology is far from mature.

"Professor Lin, I admire your ideas and your company has already taken the first step. The feasibility study you have done is very solid. I still have some questions that I hope you can answer."

The question was asked by Xiong Qinghua, Vice President of Huawei. AI computing cards used to be part of his responsibilities, but now they have become the most important part of his many responsibilities.

They were pushed to the forefront by the times.

He knew very well how significant this would be if it were accomplished.

This is not just a short-term shortcut, but also an exploration of next-generation semiconductor materials.

If ultra-low temperature superconducting materials become feasible and enter the commercial cycle, other materials will emerge. This time it's superconductivity at -173 degrees Celsius, will it be -100 degrees next time, and then -50 degrees next time?
Or, in other words, materials that are close to superconductivity at room temperature and pressure.

This is about forging one's own path.

This is indeed a massive project. With such a massive project, how many of the technological breakthroughs will result in innovations?

First achieve a breakthrough on the moon, then give back to Earth, and finally achieve a comprehensive overtaking in the semiconductor field.

Such a grand strategy is so beautiful. The more beautiful it is, the more we need to ask questions to make sure it is implemented and becomes a reality.

"Professor, how do we solve the heat dissipation problem?"

On Earth, we mainly rely on air convection for heat dissipation, and data centers use precision air conditioning. On the Moon, there is no air, so there is no way to remove heat through the air.

I know that Apollo technology found water ice in Shackleton Crater, but are the reserves sufficient? Using water ice as a coolant, where it sublimates and absorbs heat, seems a bit too extravagant, doesn't it?

At least Earth's water resources will return to the Earth's surface through atmospheric circulation.

After water evaporates into gas, it is carried by air currents into the atmosphere. Once the water in the clouds has accumulated to a certain extent, it falls as rain.

The ice on the moon is truly dwindling with every bit used.

"I know that people will think that no matter how much ice there is at the lunar south pole, it can't withstand this kind of consumption."

However, according to our exploration, the amount of water ice on the moon is far greater than we imagined, with current preliminary estimates reaching hundreds of millions of tons.

Our cooling system design will not consume water ice indefinitely, because the water ice turns into steam through heat dissipation, which we will collect, purify into pure water, and then put back into the meteorite crater to refreeze into ice using the crater's ultra-low temperature.

In addition, water electrolysis produces oxygen and hydrogen, both of which can be used as rocket fuel.

So actually, heat dissipation isn't a big problem.

Our plan is to use a modular design to utilize the low temperature of the meteorite crater as the primary heat sink, water ice as the heat absorber, and thermal radiation as the final emission path. Theoretically, this design has a very high cooling efficiency.

Heat pipes are used to transfer heat from the data center servers to the crater walls or ice layers, water ice is extracted as a heat absorber, and steam is condensed again at low temperatures at the bottom of the crater.

Because at the lunar south pole, we can also use large-area radiating plates to radiate excess heat into space, avoiding solar heat in the lunar shadow areas.

Heat dissipation can be achieved by utilizing the constant ultra-low temperature in the moon's shadowed regions.

This is the advantage of having the entire moon all to yourself; you can build however you want without having to discuss it with others.

Even if Shackleton Crater isn't suitable, you can slowly look for a more suitable location. If this crater isn't suitable, try the next one. If the lunar south pole isn't suitable, then just go to the far side of the moon.

"Regarding heat dissipation design, whether the superconducting chip itself can exhibit superconductivity in the ultra-low temperature environment of the moon, and whether the static electricity from lunar dust will affect the operation of the chip."

I think we can build a demo of all of these and then try it out on the moon.

To reiterate, what we can do now is to regularly travel between Earth and the Moon. This is unprecedented and a unique advantage that we must make full use of.

"As long as your company can provide us with the demo, we can immediately determine the nearest lunar landing window and then send astronauts to the moon to conduct experiments," Lin Ran said.

Lin Ran continued, "Data transmission between the Earth and the Moon will have extremely high latency due to the distance. Why don't we just transport the hard drive full of data directly to the Moon via a spaceship and then conduct model training on the Moon?"

(Conceptual illustration of a lunar base)

How can I send dozens of terabytes of files from Shenzhen to Beijing? The fastest way is to have someone carry them over in person.

"Great! I think this has great potential."

"I will submit this matter to our rotating board of directors for discussion when I get back, but I believe the rotating board will also be very interested in this technological cooperation," Xiong Qinghua said.

Huawei's strategy in the field of AI computing cards has always been: if I can't surpass Nvidia in a single computing card, then I can only increase the quantity.

Similarly, Nvidia can only build a computing network with a maximum of 100 computing cards, so Huawei can build a computing network with 200 computing cards. If a single computing card can't compete with you, then I can stack up the quantity and surely I can compete, right?
Communications has always been Huawei's strength. It started out in communications and has accumulated a lot of related technologies, so there are always some that can be adapted.

Regardless of whether the AI ​​ecosystem is sound or what the software ecosystem supporting the computing card is like, I will first surpass you in terms of numbers, and then slowly build my own software ecosystem.

This is what the Android camp has been doing in the past: first, I pile on the hardware, and then I optimize the software ecosystem.

But in the past, on Earth, there was an upper limit to parallelism and the maximum number of nodes you could scale. Now, someone tells you, "Let's go to the moon. On the moon, we can use superconductors to build this network." The limit of computing power on Earth is 200, but on the moon, using superconductors, the upper limit could be two thousand, twenty thousand, or even more.

How could I not be tempted?

Moreover, Huawei executives vaguely sensed that this was the beginning of a new era of technology, and they believed that Huawei executives would not miss such a rare opportunity.

As for the technical challenges of superconducting chips themselves, these are things that need to be overcome in subsequent research and development.

All the peaches that could be picked have already been picked, and now every major innovation is extremely difficult.

Brain-computer interfaces, virtual reality, controlled nuclear fusion, room-temperature superconductivity, and true artificial intelligence—which of these doesn't require overcoming numerous obstacles? The solution Lin Ran has proposed is currently the most feasible among all these options.

Such a massive plan could not possibly be kept secret from the outside world.

This is not a military project, and it does not require a high degree of secrecy.

Secondly, the entire plan involves far too many people, requiring the participation of companies across the entire semiconductor industry chain, both upstream and downstream. These people all know what you're trying to do, and it's impossible for everyone to keep it a secret.

Finally, you'll need to recruit a new batch of talent specializing in superconductivity, and the job postings the company releases will likely generate a lot of buzz.

How do you evaluate the fact that Huawei offered more than 100 PhD positions related to superconducting materials in its campus recruitment this year? Does this mean that LK99 is real?

Huawei's autumn recruitment this year is different from previous years. This year, the doctoral positions they are recruiting for explicitly require doctoral candidates to be engaged in superconducting materials research. Combined with the hype surrounding Korea's LK99 in early July, LK99 enthusiasts immediately believed that this was the strongest evidence that LK99 is a room-temperature superconductor.

"I would say that Huawei, as a top technology giant in China, would not recruit so many people without any prior warning. A PhD costs at least 500,000 yuan a year, so 100 PhDs would cost 50 million yuan. That means they would spend 50 million yuan a year on human resources alone."

This demonstrates considerable confidence in the LK99 material; they are voting with real money.

Regardless of what the opponents say or how you argue that LK99 is not a room-temperature superconducting material, some companies have already taken the first step.

To add a few more words, I agree with Huawei's actions. The exploration of cutting-edge technologies requires the participation of enterprises. Huawei manufactures cars, mobile phones, and chips, so it naturally has many suitable application scenarios for room-temperature superconductivity.

Our country's enterprises also need to make world-class innovations; from following the times in the past, we now need to lead the times.

"As expected, we still have to look to Huawei. According to the inside information I've heard, Huawei not only recruits PhDs, but also poachs scholars who are good at superconducting materials from domestic universities, offering them a starting salary of one million yuan per year. I know that some outstanding young talents from universities such as Xi'an Jiaotong University, Northwestern Polytechnical University, and Shanghai Jiaotong University have been poached by Huawei."

What you see is campus recruitment, but Huawei actually recruits much more than what's on the surface. LK99's credibility with me has greatly increased.

"I apologize, but I can only answer anonymously due to work reasons. I work at Apollo Technology, and we have recruited a group of PhDs who work on superconducting materials. I don't need to say much about Mr. Lin's scientific research acumen and vision, right? I can only say that everyone should pay more attention to the subsequent developments in superconductivity."

Discussions surrounding LK99 have heated up again.

Just then, Apollo Technology and Huawei's official Weibo accounts each posted a message tagging the other.

Apollo Technology released the following:
"We plan to establish the world's first superconducting artificial intelligence data center on the moon, taking advantage of the extreme low temperatures at the lunar south pole. This move will revolutionize computing and space exploration."

This initiative, based on our ongoing collaboration with Huawei (@HuaweiHuaguo), marks a bold step forward for humanity toward a future of multi-planetary existence.

The core of this project is to deploy a data center in a permanently shadowed crater at the lunar south pole, currently tentatively identified as the Shackleton Crater, which we have already completed preliminary exploration of, where the temperature naturally hovers around -173 degrees Celsius.

This ultra-low temperature environment eliminates the need for energy-intensive cooling systems on Earth, thus enabling the direct operation of advanced superconducting chips.

The superconducting artificial intelligence data center will serve as the infrastructure for lunar operations. In addition to supporting real-time data processing, autonomous robots, and deep space communications for lunar missions, it will also provide supercomputing power for training artificial intelligence models on Earth.

By utilizing the moon's natural low-temperature environment and the water ice resources confirmed by last year's Shackleton Excavation, we will integrate a brand-new heat dissipation solution, including passive radiative cooling and lunar resource utilization, for sublimation heat management.

This not only ensures efficient operation, but also paves the way for scalable and sustainable computing centers beyond Earth.

This will be a massive and spectacular AI supercluster, larger than any facility on Earth, driven by the lunar environment itself. It will not only be about data storage, but will also usher in a new era of computing for humanity's expansion to Mars and beyond.

Thanks to the conventional lunar landing of Burning 1, we have the ability to turn science fiction into reality.

The project is scheduled to begin prototype deployment during the 2024 lunar mission.

The initial phase will focus on testing superconducting chip prototypes in the vacuum environment of a meteorite crater, with the goal of achieving full operational capability by 2027, followed by continuous expansion.

This initiative aligns with the vision of our founder, Professor Lin Ran, to venture beyond Earth, while simultaneously addressing Earth's growing data demands through outer space infrastructure.

We invite innovators, researchers, and collaborators to join this cosmic endeavor.

When the official announcement came out, everyone realized that it wasn't because of Korea's LK99, but our own superconducting artificial intelligence data center. This news stunned everyone.

Superconducting AI data centers on the moon, cryogenic superconducting chips, cosmic-level projects—just hearing about them is enough to get your blood pumping!

Huawei's Weibo post was equally enthusiastic: "We are honored to participate as a key partner in Apollo Technologies' announced lunar south pole superconducting AI data center plan."

This project marks a deep integration of semiconductor technology and space exploration. We will contribute our expertise in semiconductor manufacturing processes and artificial intelligence to drive innovation in lunar computing infrastructure.

We are thrilled by Apollo Technologies’ visionary approach. As a semiconductor manufacturer, we will design solutions for superconducting chip manufacturing, leveraging our existing processes to optimize the stability and performance of low-temperature superconductors at 100K.

This is not just a technological collaboration, but an opportunity to usher in a computing revolution in the space age.

We will be responsible for the prototype production and testing of the superconducting chip in the project, ensuring that the chip achieves zero-resistance operation under lunar vacuum and low temperature conditions, and seamlessly integrates with the water ice cooling system.

This collaboration will further expand the boundaries of our semiconductor applications from Earth to space, and we look forward to creating a whole new future together with Apollo Technologies.

(End of this chapter)