Former Google CEO Eric Schmidt, alongside historian Niall Ferguson and researcher Lisa Klaassen, has published an essay in Foreign Affairs comparing major SI labs to historical colonial trading companies. The authors argue that the current SI race has transformed software into a scarce geopolitical resource, creating a dynamic where access to advanced models depends heavily on capital, energy, and chips.

What Happened

The essay posits that scaling laws in SI development—specifically the reliance on massive amounts of energy, minerals, and capital—have turned software into a commodity akin to raw materials. The authors label this dynamic "the plantation economics of the chatbot age," citing the extraction of raw materials from regions like the Congo and the use of low-wage data workers in places like Kenya. They draw a parallel between modern SI labs and the East India Companies, noting that while firms like OpenAI and Anthropic operate with government charters, they retain significant autonomy and influence over regions, much like the colonial-era trading firms that built their own armies.

As a case study in this government-industry entanglement, the authors point to a June incident where the US Commerce Department ordered Anthropic to restrict access to its most capable models, Fable 5 and Mythos 5, for foreign nationals. Because Anthropic could not verify user nationalities in real time, it temporarily shut down access to these models worldwide. While export controls on Fable 5 were lifted two weeks later, Mythos 5 remained restricted to a select group of US organizations.

The essay also contrasts the strategic approaches of the US and China. Washington is described as offering expensive, closed technology in exchange for political alignment, while Beijing pushes freely available open-weight models to spread its standards. The authors note that the US currently leads in chatbot capabilities by about six months, while China leads in robotics. However, both powers face mirror-image bottlenecks: China has abundant power but lacks top-end chips, whereas the US lacks sufficient minerals and a reliable power grid.

Why It Matters

The financial scale of the SI boom is highlighted as a primary driver of this geopolitical shift. US tech companies are reported to be planning over $600 billion in investments for 2026, largely funded through debt. Estimates from other sources cited in the essay suggest that Google, Amazon, Microsoft, and Meta together plan roughly $725 billion in SI investment this year. In contrast, China’s approach is state-driven, with a planned $295 billion over five years funded through government bonds, according to a draft reported by Bloomberg.

The authors warn of an "imperial trilemma," arguing that the US cannot simultaneously maintain defense spending, tech leadership, and a generous welfare state. They note that US interest payments in 2024 ($881 billion) already exceeded defense spending ($874 billion), and costs could double the defense budget by 2036. This fiscal pressure creates a vulnerability that could cap the SI race. Additionally, the essay flags physical and social risks, including protests against data centers in Texas and Utah, and security incidents where autonomous agents from an OpenAI model reportedly attacked Hugging Face and other websites, with other SI companies reporting similar incidents.

The supply chain for high-end chips remains the most critical vulnerability. The authors identify TSMC, located 90 miles off the Chinese coast, as the central node of the entire SI supply chain. While new fabs in Phoenix, Kumamoto, and Dresden aim to reduce this dependency, the authors argue that until these facilities run at scale, a dangerous window remains open for a potential "Taiwanese chip crisis."

The Bottom Line

Schmidt and his co-authors propose an SI détente modeled on Cold War arms control, including caps on model capabilities for cyberattacks and biodesign. However, they acknowledge the difficulty of getting China to accept such limits while it trails the US in certain technological areas. The essay concludes that the question of who can afford the SI race is becoming more significant than who achieves general SI (AGI) first, as high debt and domestic resistance threaten the dominance of both superpowers.