The space industry is becoming the next focus of competition. A recent close-approach incident involving satellites has drawn attention— a commercial satellite and a satellite from a leading space company narrowly missed each other within about 200 meters. The subsequent statements from both sides differ: one claims the other did not coordinate collision avoidance in advance, while the other emphasizes that the launch window was strictly selected according to ground sensing system procedures. Behind this is the reality of increasingly scarce space orbital resources.

Currently, the total number of satellites in orbit worldwide has reached 15,621. Among them, a company under a major exchange holds an absolute advantage, with approximately 10,490 satellites—almost two-thirds. In comparison, domestic commercial space has started later but is developing rapidly—currently over 950 satellites in orbit, ranking third globally, though still about 11 times behind the leader.

Why do close-approach incidents happen so frequently? The core reason is that some satellites need to frequently change orbits. In just half a year from late 2024 to mid-2025, there have been 140,000 maneuver operations—averaging one orbit change every 9 days per satellite. The technical reason behind this is that lightweight satellites with large-area solar sails are extremely sensitive to atmospheric drag, causing their orbits to decay continuously, thus requiring constant maneuvers to maintain stability.

This frequent orbit-changing practice has previously caused serious safety hazards. In 2021, there were two close-approach events between satellites and the Chinese space station, threatening the safety of astronauts, leading to emergency collision avoidance measures for the domestic space station.

However, the risk of space collisions may continue to rise. The domestic Wanxing constellation plan is enormous—overall, about 40,000 satellites are planned. Among them, the GW constellation plans to launch 13,000 satellites, the Qianfan constellation 15,000, and Honghu-3 plans over 10,000. This means that from the current more than 900 satellites to 40,000 in the future, the scale of the project is staggering.

The GW constellation is the most urgent, aiming to put all 13,000 satellites into orbit by 2035. By then, space will form a highly dense satellite network, significantly increasing the probability of orbital conflicts.

Why are all parties investing so heavily in space? The answer lies in the two core applications of the satellite industry—satellite internet and space computing power.

Satellite internet in low Earth orbit (200-2000 km altitude) networks integrates with ground 5G/6G, creating a seamless terrestrial-space fusion. This mode of communication can control latency within 15-100 milliseconds, covering remote oceans, deserts, mountains, and disaster areas—places where ground base stations are difficult to reach. In the event of natural disasters or other emergencies, the strategic value of satellite internet becomes evident.

Space computing power is another dimension of imagination. Currently, the AI chip industry faces severe challenges—hardware bubbles are widely acknowledged. Data center construction costs are exorbitant, with inflated prices for chips, batteries, and upstream components, and power supply bottlenecks after deployment, while revenue growth cannot keep pace with investment.

Moving computing centers into space immediately changes the cost structure.

For example, space solar energy has a power density five times that of ground photovoltaic systems and can supply power 24/7 without interruption. Its energy cost is only one-tenth of that on the ground. Heat dissipation is also a non-issue—space temperature is minus 270°C, making liquid cooling technology unnecessary.

Looking at construction costs: building ground data centers requires land acquisition, government approval, supporting utilities, and compliance with environmental regulations—complex processes with high costs, totaling about $167 million. Space data centers are much simpler—just launching satellites, with a total cost of around $8.2 million—down 95%.

The efficiency gains are also significant. For example, in forest fire monitoring, transmitting raw data on the ground takes hours, while space computing power can improve processing speed by 90%.

Such opportunities will not be missed. Nvidia has already sent the H100 chip into space and demonstrated a viable commercial model; Google plans to launch two computing satellites in 2027; a leading space enterprise has more aggressive plans, aiming to deploy 100GW of space data centers annually over the next 4-5 years.

The recent surge in stocks related to commercial space reflects market expectations for this wave. A new era is beginning, and investment opportunities in the space industry chain are just getting started.