|
|
|
5 Ways AI is Used in Space
BY PROCUREAM RESEARCH
|
| Space used to send us pictures. Now it is starting to send us answers. |
| The largest technology companies in the world are designing spacecraft whose only job is to compute. Artificial intelligence and the space sector now depend on each other. AI makes satellites more capable, and space is becoming a place to put AI infrastructure itself. |
|
Conventional view |
2026 reality |
| Role of AI in space |
A tool applied to missions |
A customer the space sector sells to |
| Where compute lives |
On the ground |
Increasingly in orbit |
| What comes down the link |
Raw data |
Finished answers |
| The binding constraint |
Bandwidth |
Power, land, water, and heat |
|
| Here are five ways AI is used in space today. |
| 1. Data Centers in Orbit |
| AI data centers on Earth have hit physical limits. Grid interconnection queues in some markets now stretch beyond seven years, and more than 75 data center projects worth over $130 billion were delayed or cancelled in the first quarter of 2026 alone.1 Orbit offers different tradeoffs. Solar arrays in the right orbit are up to eight times more productive than the same panels on the ground, and heat radiates directly into a vacuum rather than being cooled with water.2 |
| Programs defining the field include: |
- SpaceX* unveiled its first orbital data center satellite, in June 2026, and has filed for a constellation of up to one million of them.3
- Google* is pairing its own AI chips with satellite expertise from Planet Labs*, with two prototype satellites planned by early 2027.2
- Starcloud* is furthest along in flight, having operated an Nvidia* GPU in orbit since late 2025.4
|
| We believe the near-term beneficiaries are the suppliers: the launch providers, the manufacturers that build the spacecraft, and the companies supplying power, cooling, and laser communications links. |
| 2. Satellites That Think for Themselves |
| Onboard processing has moved from experiment to baseline requirement. The driver is decision speed, not bandwidth savings. A satellite that can classify what it sees and act without waiting for a ground pass operates on a fundamentally different timeline than one that cannot.5 |
| 3. Turning Pixels into Intelligence |
| Earth observation adopted AI first and remains the most mature application, accounting for roughly 43 percent of the AI-in-space-operations market in 2026.6 The value has shifted along the chain. Selling images is a commodity business. Selling detection, change monitoring, and classification is not. |
| 4. Traffic Control for a Crowded Orbit |
| As of early 2026 there were more than 10,000 active satellites, roughly 30,000 catalogued debris objects larger than 10 centimeters, and an estimated half million smaller objects that cannot be reliably tracked from the ground.6 At that density, collision avoidance is a daily operating cost rather than an occasional event and reviewing every close-approach alert by hand does not scale. Autonomous navigation is now the fastest-growing category of AI in space operations.6 |
| 5. Defense and Space Domain Awareness |
| Neural networks are being applied to accelerate space object detection and improve sensor tasking, while onboard autonomy is critical for managing close-proximity operations.7 Programs are now designed on the assumption that communications may be degraded or denied, which means the spacecraft has to act locally. |
| Every one of these applications runs on hardware that someone has to build, launch, and operate. The Procure Space ETF® (Nasdaq: UFO) counts among its constituents many of the satellite operators, launch providers, and component suppliers doing that work. Investors looking to participate in the space sector may consider UFO, the world’s first pure-play space ETF, for their portfolios. To learn more about the Procure Space ETF®, visit www.ProcureETFs.com. |
|
|
1 Data Centre Review, “Could the future of the data centre be in space?”, https://datacentrereview.com, August 2026.
2 Data Center Dynamics, “Project Suncatcher: Google to launch TPUs into orbit with Planet Labs”, https://www.datacenterdynamics.com, 2026; and Fierce Network, “Space data centers: Starcloud, SpaceX and Project Suncatcher explained”, https://www.fierce-network.com, March 2026.
3 Data Center Dynamics, “SpaceX details AI1 satellite ‘data center,’ claims 150kW peak compute”, https://www.datacenterdynamics.com, June 2026; and Quartz, “SpaceX reveals AI1 satellite design ahead of IPO”, https://qz.com, June 2026.
4 TechCrunch, “Starcloud raises $250 million for orbital data centers as launch options dry up”, https://techcrunch.com, August 2026; and GeekWire, “Orbital AI: Seattle-area startup Starcloud hits $1.1B valuation to build space-based data centers”, https://www.geekwire.com, March 2026.
5 Aerospace America (AIAA), “Inside the space data race”, https://aerospaceamerica.aiaa.org, July 2026.
6 New Space Economy, “AI as Mission Control: How Autonomous Satellite Operations Are Changing the Ground Segment”, https://newspaceeconomy.ca, March 2026.
7 Center for Security and Emerging Technology, Georgetown University, “AI on the Edge of Space: Securing Space Superiority”, https://cset.georgetown.edu, June 2025.
|
| Important Information |
| *As of September 22nd, 2026, Space Exploration Technologies Corp (SPCX) was a 14.75% holding, Planet Labs PBC (PL) was a 3.34% holding, Alphabet Inc (GOOGL) was a 0.00% holding, NVIDIA Corp (NVDA) was a 0.00% holding, Starcloud Inc was a 0.00% holding in the Procure Space ETF® (NASDAQ: UFO). |
| For a complete list of holdings in UFO, visit: https://procureetfs.com/ufo/. Fund holdings and sector allocations are subject to change at any time and should not be considered a recommendation to buy or sell any security. |
| Please consider the Funds investment objectives, risks, and charges and expenses carefully before you invest. This and other important information is contained in the Fund’s summary prospectus and prospectus, which can be obtained by visiting procureetfs.com. Read carefully before you invest. |
| Investing involves risk. Principal loss is possible. The Fund is also subject to the following risks: Shares of any ETF are bought and sold at market price (not NAV), may trade at a discount or premium to NAV and are not individually redeemed from the funds. Brokerage commissions will reduce returns. |
| Aerospace and defense companies can be significantly affected by government aerospace and defense regulation and spending policies. The exploration of space by private industry and the harvesting of space assets is a business based in future and is witnessing new entrants into the market. Investments in the Fund will be riskier than traditional investments in established industry sectors. The Fund is considered to be concentrated in securities of companies that operate or utilize satellites which are subject to manufacturing delays, launch delays or failures, and operational and environmental risks that could limit their ability to utilize the satellites needed to deliver services to customers. Investing in foreign securities are volatile, harder to price, and less liquid than U.S. securities. Securities of small- and mid-capitalization companies may experience much more price volatility, greater spreads between their bid and ask prices and significantly lower trading volumes than securities issued by large, more established companies. The Fund is not actively managed so it would not take defensive positions in declining markets unless such positions are reflected in the underlying index. Because the Fund is “non-diversified,” it may invest a greater percentage of its assets in securities of a single issuer or fewer issuers than a diversified fund, which may expose the Fund to the risks associated with the developments affecting the issuers in which the Fund invests. Please refer to the summary prospectus for a more detailed explanation of the Funds’ principal risks. It is not possible to invest in an index. |
| UFO is distributed by Quasar Distributors LLC. |