# 隨著越來越多科技巨頭如 Google、Elon Musk 擁有的 SpaceX 以及多家新創公司投入太空領域，美國房地產服務公司 JLL 預估全球資料中心容量將於 2030 年達到 200 GW。

*aerospace · news · 2026-05-04 · The Indian Express*

## Key points

- SpaceX 目標發射多達 100 萬顆資料中心衛星至軌道，用於 AI 工作負載。
- Meta 正與合作夥伴利用來自太空的太陽能為地面資料中心供電。
- Pixxel 與 Sarvam 的 Pathfinder 衛星將直接在軌道上執行 AI 模型訓練與推論。
- Pathfinder 將在太空中處理高光譜數據，減少向地球傳輸原始數據的需求。
- 專家表示，由於發射與維護成本高昂，軌道資料中心目前在商業上尚不可行。

The announcement comes as a growing number of tech giants such as Google and Elon Musk-owned SpaceX as well as several startups look to space as a way of bypassing Earth’s power constraints. Global data centre capacity is estimated to reach 200 GW by 2030, according to American real estate services company JLL. Meanwhile, a new report from investment bank Morgan Stanley estimates India’s data centre capacity to grow six‑fold from 1.8 GW to about 10.5 GW by 2031. However, the massive energy demand for terrestrial data centres has sparked public resistance and is pushing tech companies to explore out-of-the-box solutions such as orbital data centres. Elon Musk-owned SpaceX said it aims to launch up to 1 million data-centre satellites into orbit. Recently, Meta announced partnerships with energy startups Overview Energy and Noon Energy to tap into solar energy beamed directly to the Earth from space in order to power the social media giant’s on-ground data centres. The Indian Space Research Organisation (ISRO) is also exploring next-gen satellites with on-board data processing and data storage, according to the Centre’s response in Parliament on December 11. Story continues below this ad However, the concept of orbital data centres has been dismissed by many experts as outlandish and commercially unviable because rocket launch costs would need to fall from today’s low thousands of dollars per kilogram to the low hundreds of dollars per kilogram. In addition, power is not the only cost involved in operating a data centre, and concerns have also been raised about servicing or replacing space-based GPUs that fail during AI model training. Pixxel-Sarvam partnership As part of the partnership, Pixxel said it will design, build, launch, and operate the Pathfinder satellite. The satellite will be developed at Gigapixxel, the company’s upcoming facility that will be designed to scale satellite production to up to 100 units. Pixxel did not disclose additional technical details about Pathfinder. Sarvam, on the other hand, will handle the training and inference of its language models directly in orbit. The models and inference platform will process data with no dependence on foreign cloud or ground infrastructure, as per the press release. The mission will further validate real-time AI inference and data processing in the harsh space environment. It is expected to test performance, power management, thermal constraints, and real-time data workflows under operational conditions while establishing the technical and commercial groundwork for future orbital data centre systems. Story continues below this ad “Orbital data centres open up a new frontier, where compute can be powered by abundant solar energy, operate closer to space-based data, and move beyond some of the limits faced on Earth. For Pixxel to build the next generation of space infrastructure, we have to help shape this shift, not watch it happen from the sidelines,” Pixxel CEO Awais Ahmed said in a statement. “Sarvam has been building India’s full-stack AI platform from the ground up, and partnering with Pixxel allows us to extend that sovereign stack into space. Having India-built models running in orbit aboard an India-built satellite is exactly the kind of foundational capability that the country needs to control its own intelligence infrastructure,” Sarvam CEO Pratyush Kumar said. Besides housing chips to train AI models, Pathfinder will also carry a hyperspectral imaging camera capable of capturing high-fidelity hyperspectral data. This data will be analysed directly in orbit using foundation models trained in space. “Instead of sending large volumes of raw imagery back to Earth for processing, the system can identify patterns, detect changes, and generate insights in real time. This significantly reduces the delay between data capture and decision-making, enabling faster responses across environmental monitoring, resource management, and critical infrastructure tracking,” Pixxel said.

**Companies:** Google, SpaceX, Meta, Morgan Stanley, JLL, ISRO, Pixxel, Sarvam
**Countries:** India, United States

[Read the full story on The Indian Express](https://indianexpress.com/article/technology/artificial-intelligence/pixxel-sarvam-india-first-orbital-data-centre-ai-training-10672464/)

---

Canonical: https://newsio.io/zh-TW/n/6acf71ac-731a-49d3-a9c3-3f5cd00927db/googleelon-musk-spacex-jll-2030-200-gw
Summarized by Newsio from The Indian Express. https://newsio.io/how-it-works