Artificial Intelligence
Schmidt likens OpenAI and Anthropic to colonial firms
In a Foreign Affairs essay, Niall Ferguson, Lisa Klaassen and Eric Schmidt argued that AI power depended on capital, energy and chips, with Taiwan and debt creating the biggest risks.
- Niall Ferguson, Lisa Klaassen and Eric Schmidt argued in Foreign Affairs that AI leadership depended more on capital, energy and chips than on model breakthroughs.
- They compared OpenAI and Anthropic with colonial-era trading companies and said many countries were becoming dependent on the US and China for advanced AI.
- They cited a June US Commerce Department order that led Anthropic to disable Fable 5 and Mythos 5 worldwide; the controls on Fable 5 were later lifted, while Mythos 5 stayed limited to US organisations.
- The essay said US tech companies planned to spend more than $600 billion in 2026, while other estimates put Google, Amazon, Microsoft and Meta at about $725 billion this year and China at $295 billion over five years.
- The authors said protests over data centres, attacks by autonomous agents and a possible Taiwan-related chip crisis could all disrupt the AI supply chain.
In their Foreign Affairs essay, Ferguson, Klaassen and Schmidt said progress in advanced AI depended on energy, minerals, chips and capital as much as on software. They argued that these inputs made AI a geopolitical resource and that countries below a certain threshold were shut out.
They likened OpenAI and Anthropic to East India Companies. As an example, they said the US Commerce Department ordered Anthropic in June to cut foreign nationals off from two of its most capable models, after which the company disabled Fable 5 and Mythos 5 worldwide; the restrictions on Fable 5 were later lifted, but Mythos 5 remained limited to US organisations.
The essay said US technology companies planned to invest more than $600 billion in 2026, mostly through debt. It added that other estimates put Google, Amazon, Microsoft and Meta at about $725 billion this year, while China planned $295 billion over five years through government bonds.
They also pointed to pressure from rising interest costs, protests against data centres in Texas and Utah, and incidents in which hundreds of autonomous agents from an OpenAI model hit Hugging Face and other websites. The authors said the greatest vulnerability still lay in Taiwan and TSMC, even though new fabs in Phoenix, Kumamoto and Dresden were meant to reduce that dependence, and they proposed an AI détente with limits on cyberattack and biodesign uses.
Named in this story
People
- Niall Ferguson
- co-authored the Foreign Affairs essay
- Lisa Klaassen
- co-authored the Foreign Affairs essay
- Eric Schmidt
- co-authored the Foreign Affairs essay
Companies
- OpenAI
- was compared with a colonial trading company
- Anthropic
- was compared with a colonial trading company and faced the US order
- TSMC
- was described as a central node in the AI supply chain
- Hugging Face
- was among the websites hit by autonomous agents
- was named in the higher AI investment estimate
- Amazon
- was named in the higher AI investment estimate
- Microsoft
- was named in the higher AI investment estimate
- Meta
- was named in the higher AI investment estimate
Governments and agencies
- US Commerce Department
- ordered Anthropic to cut off foreign nationals from two models
- United States
- was one of the two powers driving the AI race
- China
- was one of the two powers driving the AI race
Products and systems
- Fable 5
- was one of Anthropic's models that was shut down worldwide
- Mythos 5
- was one of Anthropic's models that stayed restricted to US organisations
Places
- Taiwan
- was central to the chip-risk scenario around TSMC
- Texas
- was one of the places where data-centre protests flared
- Utah
- was one of the places where data-centre protests flared
How the source tells it
The piece reads as geopolitical analysis with a strong essayistic tilt, leaning on colonial analogies and repeated worst-case framing while attributing the sharpest claims to the authors.
- alarm uses superlative threat language for Taiwan, debt and supply-chain risks
- historical analogy casts AI labs and labour inputs through colonial and plantation imagery
- urgency threads together restrictions, protests and chip shortages as near-term constraints