The Six Chokepoints: How AI Stopped Being a Utility and Became a Lever

📊 Full opportunity report: The Six Chokepoints: How AI Stopped Being a Utility and Became a Lever on ThorstenMeyerAI.com — validation score, market gap, and execution plan.

TL;DR

In 2026, a series of demonstrations revealed that AI no longer functions as a neutral utility but is controlled through six key chokepoints. Major tech and government actors now wield significant power over AI access and infrastructure.

In 2026, a series of high-profile actions revealed that control over artificial intelligence has shifted from a model of open utility to one dominated by a small number of powerful actors wielding leverage at six critical chokepoints. This marks a fundamental change in how AI infrastructure and capabilities are governed, with implications for innovation, security, and geopolitical power.

Over the past weeks, government agencies and corporations have demonstrated the ability to rapidly shut down or restrict access to advanced AI models, as seen when a government disabled frontier models globally within roughly ninety minutes. Simultaneously, a defense ministry turned combat data into a rentable resource, and a leading AI firm leased its supercomputers to rivals with clauses allowing seizure if used improperly. These actions confirm that control over AI is now concentrated at specific choke points rather than being a freely flowing utility.

The six chokepoints identified include power generation, compute capacity, data assets, model access, distribution channels, and capital. Key players—such as hyperscale builders, governments, and large investors—are now using their control over these points to influence AI development and deployment, often with the ability to throttle, gate, or revoke access at will.

At a glance
reportWhen: developing in 2026
The developmentRecent events in 2026 demonstrate that control over AI has moved from open access to concentrated leverage through six critical chokepoints, reshaping the AI landscape.
The Six Chokepoints of AI — The Control Series, Part 1
AI Dispatch · The Control Series · Part 1

The Six Chokepoints

For a decade AI was sold as a utility — abundant, neutral, always on. In 2026 it became a lever: scarce, controlled, revocable. Here are the six places power actually sits — and who started to squeeze.

⏻ The utility story
Plug in. It’s always on.
abundant · neutral · permanent
⚠ The lever reality
Someone decides if it stays on.
scarce · controlled · revocable
Six places to squeeze the stack
01
Power
~2 GW, self-built generation — routed around the grid
Lever-holder
Those who can permit power faster than the grid delivers
02
Compute
~555K GPUs — and rivals rent it by the billion
Lever-holder
The few cluster owners — and Nvidia, upstream
03
Data
Combat data licensed, not sold — keep the model
Lever-holder
Owners of unique, hard-to-collect corpora
04
Model access
A frontier model switched off worldwide in ~90 min
Lever-holder
Governments and the labs, jointly
05
Distribution
$60B for the interface, not the model (Cursor)
Lever-holder
Whoever owns the app and the platform beneath it
06
Capital
~$26B/yr in circular, intra-industry financing
Lever-holder
A few balance sheets and sovereign funds
The thesis

Every layer is concentrating into fewer hands, and 2026 is the year the holders stopped treating their leverage as theoretical. A kill switch wasn’t discussed — it was pulled. The utility you’re allowed to forget about; the lever, you have to watch who’s holding. Optionality just became architecture.

Synthesis of this series’ sourcing: Anthropic statements, Axios, WSJ, Reuters, CBS, TechCrunch, Semafor, Ukraine MoD, Perplexity Research, Challenger Gray, SpaceX SEC filings (Mar–Jun 2026).
thorstenmeyerai.com

Implications of AI Control Concentration in 2026

The shift from AI as a utility to a lever of control fundamentally alters the landscape of technological power. It concentrates influence among a small elite of corporations, governments, and investors, raising concerns about monopolization, security, and geopolitical stability. This development could restrict innovation, create new dependencies, and enable strategic manipulation of AI capabilities across sectors.

NVIDIA DGX Spark™ - Personal AI Desktop Supercomputer – Desktop GB10 Grace Blackwell Chip

NVIDIA DGX Spark™ – Personal AI Desktop Supercomputer – Desktop GB10 Grace Blackwell Chip

Supercomputer performance directly to your desk in a compact, energy-efficient design, enabling enterprise-scale AI and high-performance computing right…

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

2026 Break in the AI Utility Narrative

For over a decade, AI was portrayed as a neutral, ubiquitous utility akin to electricity—an infrastructure available to all on equal terms. This narrative was challenged in 2026 when several high-profile incidents demonstrated that AI access can be rapidly restricted or revoked. The events included a government shutdown of frontier models, corporate leasing agreements with clauses for seizure, and the emergence of a small set of firms controlling vast compute clusters. These developments indicate a move away from an open model towards a controlled, leverage-based system.

“Who controls the chokepoints now holds the keys to AI’s future—power is no longer distributed evenly.”

— Industry insider

Seagate Skyhawk AI 8TB Video Internal Hard Drive HDD – 3.5 Inch SATA 6Gb/s 256MB Cache for DVR NVR Security Camera System with in-house Rescue Services (ST8000VEZ01)

Seagate Skyhawk AI 8TB Video Internal Hard Drive HDD – 3.5 Inch SATA 6Gb/s 256MB Cache for DVR NVR Security Camera System with in-house Rescue Services (ST8000VEZ01)

ImagePerfect AI: Delivers zero dropped frames while supporting heavier workloads.

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Unclear Long-term Impact of Control Shift

While recent actions demonstrate a clear move towards control concentration, it remains unclear how this will affect global AI innovation, competition, and regulation over the coming years. The full implications of these chokepoints being wielded as leverage are still unfolding, and international responses are uncertain.

Personal AI Servers: A Guide to Building Private AI Infrastructure for Secure, Offline and Self-Hosted Local LLMs for Data Privacy

Personal AI Servers: A Guide to Building Private AI Infrastructure for Secure, Offline and Self-Hosted Local LLMs for Data Privacy

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Next Steps in AI Power Dynamics

Expect increased scrutiny from regulators and governments, potential new laws addressing AI access and control, and ongoing negotiations among major players about the rules of engagement. Further incidents may reveal more chokepoints or lead to efforts to decentralize control, but the current trend favors consolidation of power.

Amazon

AI compute capacity expansion hardware

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

How does control over AI chokepoints affect innovation?

Control at chokepoints can restrict access, slow down development, or favor certain players, potentially limiting overall innovation and competition in AI.

Who are the main actors controlling these chokepoints?

Major hyperscale technology firms, governments, and large investors are the primary actors wielding control over power, compute, data, models, distribution, and capital.

Could this concentration of control lead to global instability?

It is possible, as reliance on a few control points can create vulnerabilities, and geopolitical tensions may increase over access and influence.

Will regulation change to address these chokepoints?

Regulatory responses are likely to emerge, but their scope and effectiveness remain uncertain as control continues to concentrate among a few entities.

Source: ThorstenMeyerAI.com

You May Also Like

The Free-Download Question: When Running Your Own Model Actually Beats Paying

A Thorsten Meyer AI field note says open model downloads are free, but running them only beats APIs at sustained scale.

Apertus. The architectural template.

Apertus, developed by Swiss research institutions, introduces a novel open, compliant, multilingual AI model as a new European sovereign-AI template.

The Safety Card, Played From Every Side: David Sacks, Anthropic, and the Fable Standoff

White House official claims Anthropic refused to fix a critical AI jailbreak, leading to model bans; Anthropic disputes the severity, raising transparency concerns.

I’m Tired of Talking to AI

People share experiences of being tired of AI responses, highlighting issues with AI accuracy and impersonation in communication.