Shanghai Launches the "Pearl Constellation Plan": Bringing Supercomputers into Space – What’s the Real Purpose?
Hello everyone, I’m your financial journalist. Today, we’re talking about a story that sounds like science fiction, but it’s actually accelerating towards reality: Shanghai has launched the “Pearl Constellation Plan”.
In simple terms, the goal is to deploy gigawatt-level supercomputing power – a vast amount of computing power – into low Earth orbit through a network of satellites.
It might sound like something out of *Star Trek*, but there’s a very real economic and strategic behind it. To help you understand this better, I’ll break down the news into five key aspects and explain it in plain language.
---
1. What are we doing?
In one sentence: Shanghai, in collaboration with 27 leading companies, is planning to move “ground data centers” into space to create a global “space-based computing network.”
Key points:
- Who’s involved: Shanghai Dongfang TianSuan Technology (the leading company), along with the Chinese Academy of Sciences, Alibaba Cloud, iFlytek, Beren Technology, and 26 other giants.
- What’s being done: They’re building a “gigawatt-level” space-based computing constellation. This means packing thousands of supercomputers into satellites and launching them into space to form a massive “cloud brain.”
- The goal: To become the foundation for future AI, enabling data to be processed in space rather than being transmitted all the way back to Earth.
- Current status: The project is still in the “exploration” and “early engineering” phases; it won’t be available immediately, but it’s being developed in phases.
---
2. Why now? The motivations for “space computing” in China and the US are different
The news mentions that SpaceX in the US also has similar plans, but the motivations for China and the US are quite different. It’s like two people trying to climb a mountain: one because there’s no path at the base, and the other because they want to claim the summit.
- The US’s challenge is limited electricity and heat dissipation: CEO Liu Jingjing put it plainly: US data centers already consume 5% of the country’s electricity. With the explosion of AI, building data centers on Earth faces two major problems: high electricity costs and difficult heat dissipation. Therefore, the US is eager to move computing power into space because Earth can’t handle it all; space offers unlimited solar energy and a vacuum that helps with heat dissipation (although there are still challenges with vacuum cooling).
- China’s motivation is strategic positioning: China already has a well-developed infrastructure for computing power on Earth and sufficient electricity, so it doesn’t urgently need to move computing power into space for now. The “Pearl Constellation Plan” is more about strategic considerations.
- Strategic advantage: Space is the future strategic high ground. Whoever controls space computing will have control over both “energy” and “computing” in the AI era.
- Autonomy and control: This is a fully domestically developed plan, from chips to energy. It’s a precaution against potential technological blockades, ensuring that China’s AI systems can continue to function even in extreme situations.
Popular analogy: The US is like a house with overheating and high electricity bills, so they want to move the furnace to the backyard (space); China has just installed a new furnace but wants to build a backup system in the backyard to prevent competitors from taking over the territory and for safety reasons.
---
3. Technical challenges: It’s not just about throwing computers into space
Many people think that space computing means strengthening Earth-based servers and launching them into satellites. Experts from the Chinese Academy of Sciences and Beren Technology have pointed out that this is far from a simple technical solution:
- Chips can’t just withstand radiation: Space radiation can damage chips. The traditional approach of using radiation-resistant coatings is costly and slow to innovate. The current approach combines hardware and software:
- Hardware: Improving circuit design and adding communication redundancy (like having spare tires for a car, but also relying on skilled drivers to avoid problems).
- Software: Algorithms need to be fault-tolerant, able to automatically detect and correct errors or bypass damaged parts.
- From individual units to clusters: Most current in-orbit computing power is still based on individual chips, like separate computers in space. A true computing network requires “networked clusters.”
- Challenges: How to achieve high-speed communication between satellites? The industry is exploring laser communication for networking.
- Time frame: Dong Chaofeng, the head of Beren Technology, says it will take another 2-3 years to form a real cluster. Trying to run trillion-parameter models in space now is unrealistic.
Popular analogy: It’s like building a giant Lego castle in the air:
- Traditional approach: Making each Lego piece radiation-resistant.
- Current approach: The Lego pieces don’t need to be super strong, but the “glue” (communication protocols) and “building rules” (algorithm fault tolerance) must be smart enough so the castle doesn’t collapse if a piece breaks.
- Current status: A few Lego pieces have been launched, but the castle hasn’t been built yet. We need to wait for “laser glue” (laser communication) and “automatic building robots” (reusable rockets) to be ready.
---
4. Industry collaboration: 27 companies working together
The “Pearl Constellation Plan” is not a single company’s effort; it’s a large alliance with government support, industry participation, and academic research collaboration. These 27 companies cover the entire chain from chips to energy:
- Core components and chips: Beren Technology and Alibaba Cloud provide computing chips and AI algorithms. Beren focuses on making chips suitable for space, while Alibaba Cloud offers big models and cloud computing expertise.
- Power and energy: Xiecheng Solar, Jinko Solar, and Trina Solar are leading photovoltaic companies. Space’s advantage is unlimited and unobstructed solar energy; they provide high-efficiency solar panels for satellites.
- Satellite platforms: The Chinese Academy of Sciences and Shanghai Jiao Tong University’s Space Computing Laboratory are responsible for satellite design, manufacturing, and launch.
- Communication: Although specific companies aren’t named, inter-satellite communication is crucial. This involves laser and microwave communication technologies.
Why collaboration? Space computing is a highly dependent field. Even the best chips won’t work without power; sufficient power without effective communication is useless; and data without processing is meaningless. Any weak link in the chain can render the entire constellation useless. Therefore, all key players in the industry must work together.
---
5. A phased approach: No rush, step by step
The “Pearl Constellation Plan” isn’t set to be completed next year; it’s being developed in a practical, phased manner:
- First phase: Technical validation (now – soon). Launching test satellites to verify key technologies.
- Focus: Testing domestic computing chips, new algorithms, and energy systems.
- Goal: Turning concepts into practical engineering solutions.
- Second phase: Iterative development and initial deployment. Based on successful technical validation, launching a small number of satellites to improve overall performance.
- Application scenarios: Deploying AI models in space, real-time weather analysis, and intelligent weather forecasting.
- Third phase: Building a global network in low Earth orbit, reaching gigawatt-level power.
- Goal: Becoming the foundation for future AI.
Implications for the public:
- Short term (1-3 years): You’ll see more test satellite launches, but there won’t be immediate changes for ordinary users.
- Medium term (3-5 years): Specific applications may emerge, such as more accurate weather forecasts and faster remote sensing.
- Long term (more than 5 years): A true “space cloud” service may appear, and the cost of AI computing could decrease significantly due to cheaper space-based energy, though this depends on the development of reusable rockets and advanced communication technologies.
---
Summary
The “Pearl Constellation Plan” is not just a fantasy; it’s a national strategic initiative involving industry collaboration. It’s not about solving immediate computing shortages (China still has enough power on Earth) but about gaining a strategic advantage in the AI era. It’s a comprehensive effort involving chips, energy, communication, algorithms, and manufacturing. It won’t happen overnight but will require 2-3 years or more of technical development and testing.
For ordinary people, this means that our phones, cars, and even city management may increasingly rely on “space-based” computing power. Shanghai is playing a key role in becoming the core of this “space-based computing era.”