About

Interconnecting Intelligence tracks AI from demand to buildout: who is using AI and for what, and the compute, energy and chips being built to serve it.

Why four layers

Demand for AI sets how much compute gets built. That compute needs power and chips, and each depends on long lead times: grid connections, generation, fabs and packaging. A change in one layer shows up in the others, often months or years later. Following all four in one place is how you see it coming.

Where the facts come from

Public records only. That starts with SEC filings from utilities, chipmakers, data center developers and the largest buyers of compute, and will widen to grid interconnection queues, utility commission dockets, local permits and model usage data. Much of what matters is filed long before it is announced.

How each fact is checked

Every fact is extracted from its source and checked automatically before it appears here. A fact is held back unless:

• it links to the original record;
• every number in it appears in that record;
• every company it names appears in that record;
• the record itself ties it to AI, data centers or AI infrastructure.

There are no opinions, predictions or recommendations here, and nothing is investment advice. Read the source before relying on any fact.

Origin

About ten years ago this started as Extending our Minds, a collection on artificial intelligence, forecasting and cognitive breakthroughs. Interconnecting Intelligence picks up the same thread with a narrower job: the verifiable facts of how AI is being built.