Google
Google · Chip · Filed May 8, 2025 · USPTO published Nov 12, 2026 · This week's release
Google Files Patent for Steering AI Satellites With Earth's Magnetic Field
Google's filing describes a flat, solar-powered satellite full of AI chips that steers by running electric current around its edge and pushing against the magnetic field around it. A trained AI model decides how hard to push.
- Publication number
- US 2026/0356501 A1(US20260356501A1)
- Applicant
- Google LLC
- Filed · Published
- May 8, 2025 · Nov 12, 2026
- Inventors
- Blaise Aguera-Arcas, Thomas Fischbacher, Travis Roland Beals
- Status when we published
- Examiner raised objections; the company can respond
- Document
- 49 claims
- US classification
- 701/13
- Examiner
- SANTOS, AARRON EDUARDO (Art Unit 3663)
What Google's magnet-steered AI satellite does
Right now, the computers that run AI sit in data centers on the ground, and those buildings are straining land and electrical power. Google's filing sketches another home for some of that work: flat, solar-powered satellites packed with AI chips, each roughly 2 meters wide.
The patent covers the part that keeps such a satellite aimed and positioned. The satellite measures the magnetic field around it, works out where it sits relative to things like Earth or other satellites, and feeds both into a trained AI model. The model decides how much electric current to run in a loop around the satellite's edge. That current pushes against the surrounding magnetic field, nudging the satellite's tilt or path.
Why bother steering? A satellite facing the sun collects more power, can dodge debris, and can line up with its neighbors to share data.
… a first predictive model configured to compute a first control output based on the magnetic field values and the location information …
Translation: An artificial intelligence system calculates how to steer the satellite using magnetic readings and its current position.
How the satellite pushes against Earth's field
The claimed system has three parts. A first sensor (a magnetometer, which measures magnetic fields) reads the magnetic field of some outside object, such as Earth or another satellite. A second sensor (a star tracker, sun tracker, camera or GPS unit) works out where the satellite sits relative to that object.
A predictive model, which the filing describes as a machine-learning model (software that learned patterns from past sensor data), turns those two inputs into a control output. The satellite then changes its orbit, meaning its position and speed, by adjusting its magnetic dipole force, the push it gets from its own magnetic field. It does this by running current in a loop around its circumference, and that loop reacts against the outside field (the Lorentz force, the same physics behind electric motors).
The dependent claims add options:
- Reaction wheels (spinning disks that turn the craft) and a magnetorquer (an electromagnet) that bleeds off excess spin.
- A physics-based model that computes a second answer, plus rules for settling disagreements between the two.
- Solar sails and thrusters as extra ways to maneuver.
For a swarm, one satellite can use its neighbors' data to compute settings for the others, and the group's formation can shift as the planet's magnetic field strength changes.
… adjusting a magnetic dipole force vector of the satellite module based on the first control output …
Translation: The satellite changes its trajectory by altering its magnetic forces.
Why Google is looking at computers in orbit
If you never think about where AI runs, this filing is a reason to start: it frames space as a way around the land and electricity limits of ground data centers. A satellite can only do that job if it stays pointed at the sun, steers clear of debris and lines up with its neighbors, and this patent is the steering piece. The examiner's first review rejected the claims as filed, which is routine at this stage and usually followed by amendments.
For you, any payoff sits far down the road: AI services that need less power and land on Earth. The filing cites computing elements drawing about 1,000 watts each, so the energy question is real, and a solar-powered satellite that avoids collisions protects the expensive chips on board.
Google's latest chip patent application is the 25th we've tracked in our Chip topic coverage since May, a run that includes a drone pickup stand and overheating prediction.
Source: US 2026/0356501 A1 at the USPTO (PDF). Full patent text and figures are in the official publication.
Get our take in your Top Stories
Liked this breakdown? Add Patentlyze as a preferred source on Google, and our plain-English take shows up more often in your Top Stories the next time Google patent news breaks.
Add Patentlyze on Google
One click.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings · 10 sheets
10 drawing sheets from US 2026/0356501 A1 · click any drawing to enlarge
Show the other 2 sheets
More from Google
All 1,079 Google patents →
More in Chip
All 645 Chip patents →
Nvidia
Nvidia Patents a Prediction-Based Power Steering System for Data Centers
Nvidia
Nvidia Patents a Way to Stop Processors from Forgetting How Code Fits Together
Nvidia
Nvidia Patents a Way to Stop AI Chips from Choking on Their Own Math
Nvidia
No comments yet