About
AI runs on energy.
Every AI answer is energy turned into a prediction. We work to lower the cost of both, and widen access to AI.

“Ask an AI a question and billions of trained connections light up at once, schooled on almost everything humanity has written, to turn stored knowledge into a judgement: a word, a line of reasoning, a decision.
Every answer is electricity. A token is energy spent to produce a thought.
That is why making intelligence abundant takes two things together: cleaner energy to power every answer, and lead models that delegate to domain-specific ones — more accurate on the task and far lighter on compute.”
The evidence
The price of a single answer is falling fast, roughly 10× a year. But demand is rising faster, from one chatbot reply to hundreds of agents working a problem together. AI token processing grew about 330× in two years, with about 24× more forecast by 2030. Cheaper tokens are being used far faster than they get cheap.
Firing one of an AI's connections for a single word costs about five trillionths of a joule. The bill is in having billions of them, fired for every word.
Demand for AI is racing ahead of the energy and economics to power it.
- ~330×
- Token processing growth in two years
- Google, 2024 to 2026
- ~10×/yr
- Cost of AI capability falling
- a16z, Epoch AI, 2024 to 2025
- 2×+
- Data-centre electricity by 2030
- IEA, 2025
- 2,290 GW
- US power projects waiting to connect; firm power takes 5+ years to build
- LBNL 2025; GE Vernova 2026
Every problem that narrows the distance
We work on every problem that narrows the distance between what AI will demand and the energy and cost to deliver it.
Two of the biggest levers today: build abundant clean energy, so AI is cheaper to run; build efficient, domain-specific models combined with the benefits of leading LLMs, so the cost per token falls and accuracy rises.
This is visible in what we build today: Energy, Canopy and Streamline.
Building for the next 15 years
We are building for the long term. We have a plan and a framework to bring down the cost of energy and the cost of AI, and to widen access to both. Over the next 15 years, nothing across the energy and AI supply chain is off the table.
Executive leadership team
Operators, engineers and builders. People who pick up the phone.
-
Seth Thuraisingham
Founder & CEO
LinkedIn (opens in a new tab) -
Ed Cottle
CFO
LinkedIn (opens in a new tab) -
Ben Schultz
VP & CTO, Lunio Labs
LinkedIn (opens in a new tab) -
Damien Ferragamo
Head of Projects
LinkedIn (opens in a new tab) -
Rafael Kargren
CCO, Lunio Labs
LinkedIn (opens in a new tab)
Real assets. Real software. Strong governance.
Australian-owned. We deploy, or it doesn't count.
Australian-owned
Built and operated from Australia, for operators who need people who pick up the phone.
Real assets and real software
Grid-scale storage in the field and software in production. Not slideware.
Strong governance
Enterprise-grade security and the discipline that comes with operating critical infrastructure.
We deploy, or it doesn't count
Everything ships to real customers and is judged on real-world outcomes.
