# Opinion: A haiku from orbit is a milestone, not a business plan

URL: https://ai-pulse.aglaganova.com/blog/opinion-a-haiku-from-orbit-is-a-milestone-not-a-business-plan
Published: 2026-11-04
Author: Roman Aglaganov, Full-stack developer
Category: Opinion
Region: Global
Tags: energy, chips

> Google’s satellite test is real engineering progress. But data centres in space depend on launch prices that do not exist yet, and they should not become an excuse to stop fixing the grid on Earth.

![Illustration: a tiny black satellite far above a large grey planet curve, with a long dashed grey line between them and a red dot on the surface](https://ai-pulse.aglaganova.com/api/media/file/opinion-orbit-1.png)

Google’s Project Suncatcher satellite has run its Gemma 4 model on a TPU in orbit and sent back a haiku. Credit where it is due: putting a data-centre chip in space, powering it up and getting a useful answer is hard, and the team did it within weeks of launch. But the excitement around “data centres in space” is running well ahead of what this test shows.

## What it proves, and what it doesn’t

It shows that a modern AI chip can run in orbit, at least for now. It does not show that thousands of them can be kept cool, linked together at the speeds AI training needs and kept running for years, all problems Google’s own 2025 design paper lists as unsolved. Nor does it show the economics. That paper estimated that launch prices would need to fall below about $200 per kilogram, perhaps by the mid-2030s, for space-based computing to be roughly comparable with energy costs on the ground. The plan rests on a forecast, not a price anyone can buy today.

Google is honest about this; it calls Suncatcher a “cosmic moonshot”, and its next step is two linked satellites in 2027. The risk is not in Google’s research but in how it is heard. It is tempting to treat orbital data centres as the answer to the power and water that AI needs, as if the problem might soon float away.

## The problem is on the ground

It won’t, at least not this decade. The data centres being built now need electricity now, which is why companies are installing gas turbines next to them and why local opposition has become an issue in this year’s US elections. Those are the decisions that will shape AI’s footprint for the next twenty years: how fast clean power and transmission lines are built, who pays for grid upgrades, and how much pollution communities are asked to accept.

Keep funding the moonshots; some of them land. But a haiku from orbit should not take attention away from the much less glamorous work on Earth.

## Sources

- [Google: We’ve received our first response from the Project Suncatcher prototype satellite](https://blog.google/innovation-and-ai/models-and-research/google-research/project-suncatcher-satellite-response/)
- [Google Research: Exploring a space-based, scalable AI infrastructure system design](https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/)
- [TechCrunch: Judge allows NAACP lawsuit over xAI data center pollution to move forward](https://techcrunch.com/2026/10/31/judge-allows-naacp-lawsuit-over-xai-data-center-pollution-to-move-forward/)