The Invisible Trillion · Who Decides

The Quieter Rooms

Those are the rooms that produce most of the visible decisions. A few others operate in the background, less dramatically and just as durably.

There’s the standards room: the technical standards for networking, data formats, security, and interoperability set by bodies like the IEEE, IETF, ISO, and ITU, through slow consensus processes almost no one outside them follows. The people who show up to those meetings, and the companies paying them to, quietly shape what every product built on the standards can do.

There’s the research room, which may be the most powerful of all and the least structured. A breakthrough by a few dozen people can move the entire industry: AlphaFold came out of a team of around thirty, and the paper that launched the modern transformer era had eight authors. What those researchers choose to work on, to publish, to hand to industry or keep in the lab, propagates through everything downstream — and a research community that organizes around safety produces different outcomes than one that organizes around capability alone.

There’s the open-source room. Much of the stack — Linux, Python, PyTorch, thousands of other tools — is maintained by loose networks of volunteers and semi-professionals whose choices about what to build and what to abandon ripple through every company that depends on them. The rise of open-weight models — Llama, Mistral, DeepSeek, the Qwen family — is reshaping the competitive structure of AI in ways the closed providers can’t fully control.

And there’s the journalism room, which is why anyone outside the companies knows any of this. Nearly every specific fact in this book — the grid cost-shifting figures, the NDA pattern, the utilization rates, the turbine counts, the shell companies — came from investigative work, not corporate disclosure. The quality of public understanding depends on whether that work gets funded and published, which makes the collapse of local journalism — private-equity owners, ad revenue dependent on the very industries the papers should scrutinize — a mechanism in its own right, one that weakens the only layer producing visibility into the others.

Advocacy organizations belong here too, at their best and worst: the Southern Environmental Law Center won real results in Memphis by engaging the Clean Air Act’s technical details and building an actual case, while plenty of groups run on little more than indignation. The difference between the two is whether they do the legal and technical work, and the ones that do have outsized effect.

All 19 chapters of The Invisible Trillion · Library · All courses