The next major investing edge in AI may come from identifying where capital for orbital computing flows first, according to Matthew Tuttle, CEO and CIO of Tuttle Capital.
Just months ago, anything tied to OpenAI could send tech stocks soaring. Today, the mood has cooled. Investors are more skeptical, and the AI sector now faces the classic “front-cover curse” after Time named several AI leaders as Persons of the Year—including Nvidia’s Jensen Huang, who also just received Financial Times Person of the Year honors.
AI’s narrative may need a fresh spark. That spark could be space.
The Wall Street Journal reports that Jeff Bezos and Elon Musk are exploring orbital data centers designed to support Earth’s growing AI workloads.
In a recent blog post, Tuttle breaks down how investors can position for this next frontier. The biggest constraint for AI’s expansion, he argues, isn’t chips—it’s power. “AI and data centers are set to drive a major surge in electricity demand this decade, which is why ‘grid tech’ has become a real market theme,” he writes.
Space, however, has a distinct energy advantage. Orbital data centers can harness uninterrupted solar power. “That’s where ‘compute in orbit’ becomes a serious idea,” Tuttle says. “AI has extremely high revenue per kilowatt, and solar in space is always on—unlike Earth-based constraints.”
Scaling orbital computing and delivering that power back to Earth remains the main challenge. The real opportunity, Tuttle says, is recognizing where capex flows first—into edge inference and space infrastructure—while full-scale gigawatt beamed power remains a long-term, sci-fi vision.
He lays out a three-phase investable roadmap:
Phase 1: Happening Now — On-Orbit Inference
Existing satellites used for weather, communications, Earth observation, maritime tracking, and missile warning.
“If satellites classify, compress, and decide onboard, they downlink answers—not terabytes,” he writes. “That cuts bandwidth, speeds up action, and reduces the need for ground stations. That’s today’s ‘sky-brains.’”
Phase 2: Purpose-Built AI Satellites
Satellites designed with compute modules.
“You’re not just processing your own sensor stream—you’re renting compute,” Tuttle notes. Challenges include radiation tolerance, launch cadence, and networking. This works for niche workloads where latency or sovereignty matters more than cost.
Phase 3: Space-Based Solar Power
Beaming solar power from space to Earth. This requires massive lightweight structures, conversion systems, beam-safety protocols, and grid integration—making it the hardest engineering challenge.
For the first two phases, Tuttle highlights potential winners in space infrastructure: Redwire, Rocket Lab, L3Harris Technologies, RTX, Northrop Grumman, and Lockheed Martin.
“These companies benefit from any version of orbital compute or space power,” he says, “because they supply spacecraft platforms, integration, communications, and space-qualified components.”
Radiation-tolerant processors will also be crucial. Tuttle points to Microchip Technology as a direct play on space-grade silicon.
But near term, most AI workloads will stay Earth-based. The real spending boom, he says, is still in the power grid—transformers, cooling systems, and distribution. His picks for that theme include Eaton, Hubbell, Quanta Services, and Vertiv.
Asked whether any ETF captures this orbital-AI opportunity, Tuttle joked: “Not until I launch it.”