we_are_coded.by CODE · The world, decoded
БГ
Who's who

AI factory

The BasicsUpdated on 16 August 2026we are coded

A building with machines, power and cooling whose output is compute. It gets financed like a power plant, not like software.

Checked on16 August 2026
In short: when you hear someone is building AI, most often a building is being built. Inside are rows of machines, a lot of power and cooling, and the output is measured in compute hours. Because this costs billions, the money comes from the same places that finance roads and power plants, and expects the same: long horizons and predictable income.

Update, 16 August 2026: the wires are part of the factory too. On 11 August NVIDIA explained the 800 VDC architecture - high-voltage direct current so there are fewer conversions between the grid and the chip and more of the power reaches the compute. NVIDIA, Google and Microsoft develop it through the Open Compute Project; according to NVIDIA more than 80 manufacturers are already building equipment to the specification. A hybrid 800 VDC rack for existing buildings arrives in the second half of 2026, a row-level power centre up to 2 megawatts - in 2027. Models change in months; a building's electrical installation - once.

The visual is generated code art. No third-party images.
Official primary sources
→NVIDIA - Why Scaling AI Compute Performance Requires a New Power Architecture

Up, to the basics

the bigger picture
Basic

Who pays for the power

Every answer artificial intelligence gives you passes through a real building, full of machines that seriously eat power...

Down, to the specifics

the finer detail
Concept

Why chips talk to each other more than they compute

A big model isn't trained on one chip, but on thousands, constantly talking to each other, and that network between them...

News on this topic

AI factory
Infra
NVIDIA Blog16 September 2026Energy

Google, NVIDIA and Emerald AI form an alliance for data centres that can cut their own power when the grid is strained

The AI Energy Management Alliance wants flexible data centres to connect to the grid faster in exchange for a verifiable promise to give power back. A day earlier NVIDIA described what that looks like live: from 4 to 3 megawatts, automatically.

Read →
Infra
NVIDIA Blog16 August 2026Infrastructure

NVIDIA wants AI factories to move to 800 volts direct current

The bottleneck is not how much power there is but how many times you convert it on the way to the chip. NVIDIA, Google and Microsoft are writing a common standard through OCP; more than 80 manufacturers are already building to it.

Read →
Infra
NVIDIA Blog14 August 2026Money

NVIDIA is selling compute as infrastructure, not as machines

The goal is over $500 billion of outside capital flowing into AI factories the way it flows into roads and power plants. NVIDIA backstops up to 25 per cent of residual value per project.

Read →