ORIJINS · SOLAR INITIATIVE

The Sun Powers Everything. Why not intelligence?

Earth receives 173,000 terawatts of sunlight every second — enough to run all of humanity, ten thousand times over. ORIJINS Solar is the bridge between that infinite light and the AI compute that will shape our century.

173,000 TW · solar input
770M · without electricity
4.3% · grid used by AI

770 million people are still in the dark.

Meanwhile, the most powerful technology ever built — artificial intelligence — burns fossil fuels by the gigawatt. We are creating intelligence by destroying the very planet that gave rise to it. This is the contradiction at the heart of our era.

0million
In the Dark
People worldwide who still live without access to electricity — the majority across sub-Saharan Africa and South Asia. A solvable problem, hiding in plain sight.
Source: IEA, Energy Access Outlook 2025
0% of grid
AI's Carbon Bill
Share of global electricity consumed by data centers, growing roughly 30% per year. By 2030, AI alone could consume more power than the entire nation of Japan.
Source: IEA, Electricity 2025
0%
Fossil-Powered AI
Roughly 60% of global data-center electricity still comes from coal and natural gas. The smarter the model, the heavier the carbon footprint behind every token generated.
Source: Ember Climate, 2025
0g CO₂
Per AI Conversation
Average emissions of a 1,000-token chat with a frontier LLM running on the grid average. Multiply by billions of users, billions of times a day.
Estimated, third-party LCA studies

What if every thought a machine ever had
was drawn from sunlight?

— a question we refuse to stop asking.

ORIJINS Solar Compute — built for sunlight, from the silicon up.

Most "green AI" is retrofit theater — fossil-fueled clusters with offsets stapled on. We are doing the opposite. ORIJINS hardware is engineered for solar from day one: low-watt, edge-located, batteries first, grid second.

Sun-Native Hardware

Mac Studio M-series and NVIDIA DGX Spark clusters tuned for low-wattage inference. Up to 11× more tokens per kilowatt-hour than legacy GPU farms.

// 280W per cluster · 24V DC native

Local-First Compute

GAIA runs at the edge — closer to the panel, closer to you. No transcontinental data center round-trips means less energy, more privacy, faster intelligence.

// edge inference · 47ms p50 latency

Solar-First Power

Each ORIJINS node ships with rooftop-ready PV interfaces, LiFePO4 storage, and load-shifting firmware that schedules training when the sun is highest.

// 72h reserve · DC microgrid ready

Carbon-Aware Routing

The GAIA scheduler routes workloads to whichever ORIJINS node is currently running on the cleanest electrons — Sahara at noon, Atacama at dusk, Tokyo at sunrise.

// gCO₂/token, real-time optimized

Returns More Than It Takes

Every flagship ORIJINS site is designed to feed surplus solar back to the local grid — an AI data center that produces, on net, more clean energy than it consumes.

// net-positive by design

Energy Access by Default

Every ORIJINS micro-cluster deployed in an underserved region brings grid-equivalent solar capacity along with it. AI that arrives with the lights, not after.

// compute + electrification, paired

173,000 TW vs. 18 TW.

The sun delivers more energy to Earth in one hour than humanity uses in one year. The problem has never been supply. The problem has been will.

Solar energy reaching Earthcontinuous, every second of every day
173,000 TW
100% — the upper bound of what's possible
All of humanity's energy use— power, transport, industry, everything
18 TW
0.0104% — what we actually consume
Global AI compute energy— all data centers, training + inference, 2025
~0.05 TW
0.000029% — a rounding error of the sun's gift

We could power every AI ever built — every model, every datacenter, every inference, forever — with less than three thousandths of one percent of the sunlight already striking the Earth. The future is not constrained by physics. It is constrained by imagination.

By 2050, every ORIJINS computation will be solar-powered.

A roadmap, not a promise — measured in deployments, megawatts, and lives changed. We publish progress quarterly. We will be measured by it.

2026 · Now
Solar-ready local compute clusters
Mac Studio M4 Ultra and DGX Spark reference designs ship with 24V DC input, LiFePO4 battery rails, and rooftop-PV connectors. First GAIA inference runs on amateur-deployed solar in Dakar, Tokyo, and California.
2028
Three flagship solar compute campuses
The first ORIJINS Solar campuses go online in the Sahel, Atacama, and Mojave — each pairing 50 MW of solar with on-site GAIA compute, plus surplus electrification for surrounding communities.
2032
Carbon-aware routing across the network
All GAIA workloads dynamically migrate between ORIJINS solar nodes globally, following the daylight. Verified gCO₂-per-token metrics published in real time, on-chain and open to audit.
2040
100 million solar-paired users
100 million households across underserved regions using GAIA on rooftop-solar microgrids deployed in partnership with ORIJINS Solar. Compute and electricity, arriving together — not as charity, but as infrastructure.
2050
100% solar. Net-positive globally.
Every ORIJINS computation — training, inference, archival — runs on solar. The network as a whole returns more clean energy to the grid than it consumes. Intelligence, finally, in balance with the planet.

Power the future with sunlight.

Engineers, energy operators, landowners, researchers, dreamers — if any part of this story called to you, write your name on the wall. We're publishing quarterly progress, and you'll see it first.

No spam. Quarterly only. Powered by the sun.