MILOS Duncko

Dealmaker, Vision Imagineer who believes in turning ideas into reality. Energy Disruption focus. Private Equity Investment, Materials & Steel expert.

[MILOS Duncko @MilosDuncko]

Intelligence was the scarcest resource in human history. AI allows us to produce it artificially, but its production is now limited by energy — by how efficiently we convert electrons into intelligence tokens, and by how much energy we can access.

[MILOS Duncko @MilosDuncko]

Energy + Compute, Infrastructure, Models + Applications are the practical unfolding of how conversion actually happens.
Layer 1 — Energy: The raw input, the fundamental fuel. ⚡
Layers 2, 3, 4 — Conversion: Where energy becomes intelligence (compute, infrastructure, models) 💻🛠️🤖
Layer 5 — Applications: The output. Where humans actually use and consume artificial intelligence. 👥

[MILOS Duncko @MilosDuncko]

AI is intelligence outside the human brain — created by converting energy into the form of knowledge.

[MILOS Duncko @MilosDuncko]

AI = Energy

AI is intelligence outside the human brain, created from energy, and existing in the form of knowledge.

[MILOS Duncko @MilosDuncko]

>cheap energy = flourishing individual
>energy = more better, every time, all the time

More energy provided for everybody on cheapest cost, everybody would be better on the long run.

Your well being is directly related to how much energy you can use or available to you inexpensive.

[MILOS Duncko @MilosDuncko]

Cheap energy = human flourishing.

More abundant energy means more prosperity, better health, more opportunity — every single time.

The cheaper and more accessible energy is, the better life gets for everyone.

[MILOS Duncko @MilosDuncko]

Civilization advances when energy becomes cheaper and more abundant.

Human well-being is directly tied to access to inexpensive energy.

More energy for more people = a better future.

[MILOS Duncko @MilosDuncko]

Energy abundance is prosperity.

The more cheap energy people have access to, the richer, healthier, and freer they become.

[MILOS Duncko @MilosDuncko]

Congratulations!

[Andrej Karpathy @karpathy]

Personal update: I've joined Anthropic. I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D. I remain deeply passionate about education and plan to resume my work on it in time.

[MILOS Duncko @MilosDuncko]

Human flourishing scales with energy abundance.

When energy is cheap, reliable, and widely available, people live better lives.

Nearly every form of prosperity is downstream of affordable energy.

[MILOS Duncko @MilosDuncko]

Electrons are everywhere — embedded in matter.

The next civilizational shift is not discovering new energy sources, but architecting how electrons move.

By organizing electron flow through intelligent, decentralized systems, energy becomes controllable infrastructure rather than static supply.

[MILOS Duncko @MilosDuncko]

For nearly 2 million years, fire - the Sun’s first face - was our harnessed echo of hēlektor’s power. These three — fire, amber, and the Sun — represent the ancient progression of energy: fire, the Sun’s raw power; amber, the Sun’s energy in electron form; and the Sun itself, the ultimate source of all energy.

[MILOS Duncko @MilosDuncko]

ēFLØW is Mastering Electron Flow for the Intelligence Era.

Liberating Electrons for the Age of Intelligence.

[ēFLOW @eflow_e]

ēFLØW: Powering the AI Economy.

We don’t just deliver energy - we build systems that move electrons intelligently, efficiently, and automatically.

We power the AI factories.

In this new era, electrons are the fuel of intelligence - and ēFlow is how they flow.

[MILOS Duncko @MilosDuncko]

Freedom is the quiet strength to walk away from almost anything - not because you lack feeling, but because you’re no longer enslaved by craving.

It’s remaining sovereign over yourself in a world that constantly tries to purchase pieces of you.

The less you need, the freer you become.

[MILOS Duncko @MilosDuncko]

ēTwin is one of ēFLØW’s core capabilities and the foundation of ēSystem — a next-generation electrical system purpose-built for AI factories. Every asset, workflow, and connection is modeled in a live digital design, allowing AI to simulate millions of scenarios before anything is built.

[ēFLOW @eflow_e]

ēTwin lets us build power systems entirely digitally before anything is constructed. A live digital twin models every asset and connection, while AI simulates millions of real-world scenarios in hours — from grid failures to market shifts — continuously improving with live data.

[MILOS Duncko @MilosDuncko]

>cheap energy = flourishing individual
>energy = more better, every time, all the time

More energy provided for everybody on cheapest cost, everybody would be better on the long run.

Your well being is directly related to how much energy you can use or available to you inexpensive.

[MILOS Duncko @MilosDuncko]

OpenAI does not own the Abilene data center. The facility was developed by Crusoe in partnership with Blue Owl and Primary Digital. It is leased and operated by Oracle.

[Shanu Mathew @ShanuMathew93]

Where am I getting my 20-30GW a year !!

Barclays (Thompson): "Assuming $50bn per 1 GW of compute, ~$670 bn of 2026 hypercale capex implies nearly 13.5 GW of compute additions. AMZN added 3.9 GW in 2025, including 1.2 GW in 4Q25, with capacity up 2x vs. 2022 and reiterated plans to 2x again by 2027, plus noted it will "add a lot more in 2026 and 2027 and in 2028 for that matter." MSFT added ~1 GW in F2Q26 alone vs. 2+ GW in F2025. OpenAI disclosed it grew from 0.2 GW in 2023 to 0.6 GW in 2024 and ~1.9 GW in 2025.1 xAI is approaching ~2 GW"

[MILOS Duncko @MilosDuncko]

OpenAI does not own the Abilene data center. The facility was developed by Crusoe in partnership with Blue Owl and Primary Digital. It is leased and operated by Oracle.

[Shanu Mathew @ShanuMathew93]

@MilosDuncko Abilene

[MILOS Duncko @MilosDuncko]

I think OpenAI doesn’t own any AI Factory yet. Am I wrong?

[Shanu Mathew @ShanuMathew93]

Where am I getting my 20-30GW a year !!

Barclays (Thompson): "Assuming $50bn per 1 GW of compute, ~$670 bn of 2026 hypercale capex implies nearly 13.5 GW of compute additions. AMZN added 3.9 GW in 2025, including 1.2 GW in 4Q25, with capacity up 2x vs. 2022 and reiterated plans to 2x again by 2027, plus noted it will "add a lot more in 2026 and 2027 and in 2028 for that matter." MSFT added ~1 GW in F2Q26 alone vs. 2+ GW in F2025. OpenAI disclosed it grew from 0.2 GW in 2023 to 0.6 GW in 2024 and ~1.9 GW in 2025.1 xAI is approaching ~2 GW"

[MILOS Duncko @MilosDuncko]

xAI’s Colossus 2 and Tesla’s Cortex 2 represent the fastest-built and among the largest AI compute factories ever constructed.

Colossus 2 is the world’s first gigawatt-scale AI cluster. Cortex 2 is rapidly expanding as a major AI supercluster.

These showcase unprecedented speed in building massive, power-hungry AI Factories — no one else comes close.

[Shanu Mathew @ShanuMathew93]

Again my Q continues to be how are we going to be pushing 20-30GW/year in aggregate if the below is true for one of the fastest & most aggressive scalers in AWS

“Outside these two [sites from AWS], no other site is moving fast enough for a GW-scale delivery in late 2026 or 2027.”

[MILOS Duncko @MilosDuncko]

xAI’s Colossus 2 and Tesla’s Cortex 2 are fastest built largest AI Factories.

[SemiAnalysis @SemiAnalysis_]

Outside these two, no other site is moving fast enough for a GW-scale delivery in late 2026 or 2027. AWS has broken ground on some GW-scale sites in Georgia, North Carolina, and Pennsylvania, but progress there is slower. Big picture: AWS’s plan to double power capacity by 2027 is back-end loaded. Indiana, Mississippi, and international expansion will define whether the next phase of growth hits on time, and it looks like it could be a major acceleration in 2027. (2/2)