Q1 · 2027 · FALCON 9
Odysseus-1.
Our first vehicle. A single orbital supercomputer that will run real workloads for real customers within weeks of reaching orbit.
The opening chapter of a longer odyssey.
EL SEGUNDO, CALIFORNIA
ODYSSEY
Compute without Earth's limitations.
A new form of compute in space, made for space
BEGIN
2027The first flight.Odysseus‑1 reaches orbit. One Odyssey supercomputer, our flagship vehicle, advancing the future of compute in space and running commercial AI workloads in LEO.
2027 · Q4The first fleet.The first fleet of Odyssey supercomputers. Linked by laser. Operating as a single distributed system.
2029The constellation.The first constellation of Odyssey supercomputers arranged across multiple orbital planes. Continuous coverage of every longitude on Earth.
2030A layer above the world.The Odyssey fleet scales into the hundreds. A continuous compute layer above the atmosphere — always in sunlight, always on.
2031Lunar orbit.Odyssey reaches the Moon. A lunar constellation joins the terrestrial fleet — two worlds, one network, bridged by light.
2033And Mars.Odyssey reaches Martian orbit. Three worlds encircled, each with its own supercomputer fleet.
2035 and beyondAn infrastructure built for a multi-planetary future.
2027
01
Mission
Civilizational progress was always limited by materials. Today, AI is limited by our planet.
Earth's grids and climate can no longer sustain the compute we need.
The solution isn't exporting Earth's infrastructure. It's building a new one.
We are building a fundamentally new compute engine. A space-native architecture designed to leverage the physics of orbit to deliver infinite scale.
2030 · PROJECTED
The grid will run out first.
Data-center electricity demand is climbing past every utility's ability to expand. Interconnect queues stretch past a decade. Transformers are backordered into the 2030s. Every hyperscaler is racing for the same scarce watts.
TODAY
The water is almost gone.
Every major AI site draws millions of gallons from freshwater basins that are already strained. Communities are pushing back. Governors are intervening.
18 TO 36 MONTHS
Building new infrastructure to meet the exploding demand takes too long.
A new data center is built on a multi-year timeline. Frontier AI moves on a much faster one. On Earth, compute infrastructure is always chasing the curve.
Odyssey specializes in building and launching new forms of compute that are made to thrive in space.
02
A New Class of Capabilities
WHAT OUR FLEET DELIVERS
01
A Capacity that Grows at an Exponential Rate
Every Odysseus supercomputer is equipped with an ultra-high-density, energy-efficient, space-optimized intelligence engine along with multiple ISLs operating as a node in the larger constellation-scale system.
02
A New Approach to Thermal Efficiency & Radiation Tolerance
Odyssey's new compute architecture decouples compute density from heat generation. We do not fight entropy; we harness thermal and radiation-induced noise as a computational resource.
03
A Model for Orbital Compute that Offers Fast, Low-Cost Compute from Day 0
Because of the ultra-high efficiency and capacity per node, running workloads on an Odysseus supercomputer costs order(s) of magnitude lower than the lowest tier on an Earth-or-Space-based traditional computer.
WORKLOADS
Compute that delivers the most efficient conversion rate from Watts to Intelligence at the most efficient price, in orbit
01
Frontier AI.
Training and inference for the largest models, at the scale the next generation of AI labs will need. The work that the grid cannot accommodate.
02
Real-time Earth intelligence.
Satellite data analyzed in orbit, where it is captured. Results delivered in under a minute. No raw downlink. No ground queue.
03
Scientific discovery.
Climate, genomics, astronomy, materials. Workloads that require a decade of Earth-based expansion can begin in orbit within days.
04
Sovereign infrastructure.
A compute layer outside any single grid, any single jurisdiction, any single point of political or environmental failure. Intelligence that cannot be switched off.
03
Our Vehicles
TIME TO FIRST FLIGHT
—
Days
—
Hours
—
Minutes
ODYSSEUS VEHICLE CLASSES
A high-density, high-bandwidth laser-linked constellation in LEO
Three vehicle classes. One mesh. Each vehicle equipped with an ultra-efficient compute engine
S-Class · Scale-Out Worker
S·
>200PFLOPS
<200 kg
Edge inference, scale-out sampling, EO processing.
M-Class · Performance Core
M·
>400PFLOPS
<400 kg
Core training
L-Class · Heavy Lifting
L·
>1,000PFLOPS
<600 kg
MoE routing, gradient aggregation, checkpoint storage.
04
Reserve
04 — RESERVE
Reserve your compute.
Early compute allocation on Odysseus-1 and the fleet that follows is opening now. Researchers, operators, builders — tell us what you need, and we will meet you at the launch window.