America, China, and the Costs of the Bipolar AI Race
The competition between Washington and Beijing over AI is intensifying. The rest of the world has little say in how this technology is built, tested, and governed.
In the era of artificial intelligence, the United States and China have emerged as the two dominant players, leading to a bipolar global AI landscape. This exclusive control over AI technologies has created significant geographic divides, with most countries relying on AI developed by these two nations without a say in their creation and governance.
As the great power rivalry between the U.S. and China escalates, other countries are increasingly dependent on AI systems, but decisions regarding their safety and control are largely made elsewhere.
The concentration of AI power has intensified the sense of vulnerability outside the U.S. and China, prompting various nations to strive for sovereign AI models of their own. For instance, the U.S. received approximately 72% of the world's new AI computing capacity between 2022 and June 2026, while China accounted for about 10%.
This disparity extends to other aspects of AI development, such as data centers, where the U.S. and China together accounted for around 70% of the electricity consumed by data centers worldwide in 2024, leaving only 30% for the rest of the world.
Furthermore, the training data and language coverage for AI are predominantly English-based, leaving most of the world's 7,000 languages poorly supported. In fact, only about 0.2% of the data used to train frontier models is in Hindi, a language spoken by over half a billion people. AI-driven applications can also lead to cultural and linguistic homogenization while perpetuating biases such as gender, disability, caste, and religious biases.
The cost of using AI can also vary depending on the language. AI companies often charge based on tokens, and the same message may require more tokens in Hindi and Odia compared to English. This "language tax" can affect developers building AI-driven apps in non-English languages, as they may pay more to process the same content than they would in better-supported languages like English, Mandarin, or Spanish.
Middle powers, such as India, Canada, France, and the United Arab Emirates, are actively developing their own sovereign AI models to gain more control over the technology and protect their national interests. These efforts are driven by concerns over access to American and Chinese frontier models, which can be restricted by governments citing national security concerns. As the rest of the world navigates this bipolar AI race, their future largely depends on the actions of the U.S. and China.
Rewritten by urgent.news; may contain errors — check the original.