Global AI chip oligopoly fractures as China’s state-backed firms bypass Nvidia via Huawei’s indigenous alternatives
Original framing: “Huawei’s AI chip sales surge as Nvidia stalls in China” — Financial Times
The current AI chip rivalry is the latest chapter in a century-long struggle for technological hegemony, from the vacuum tube wars of the 1950s to the semiconductor trade wars of the 1980s. China’s push for indigenous AI chips echoes Japan’s VLSI project in the 1970s and South Korea’s chaebol-driven semiconductor ascent in the 1990s, both of which were state-orchestrated. The U.S.-China bifurcation also recalls the Cold War’s dual-use technology splits, where military and civilian tech were indistinguishable, now repurposed for AI-driven surveillance and control.
The surge in Huawei’s AI chip sales is not merely a market correction but a symptom of a deeper tectonic shift: the fragmentation of the global AI supply chain into state-aligned blocs, each pursuing technological sovereignty at the expense of collaboration.
This realignment is the culmination of decades of U.S. dominance in semiconductors, China’s state-led industrial policy, and the weaponization of technology in geopolitical contests, from the 2018 ZTE ban to the 2022 CHIPS Act. The bifurcation risks entrenching surveillance capitalism in China while accelerating the militarization of AI in the West, all under the guise of 'national security.' Indigenous and marginalized voices are systematically excluded from these decisions, despite bearing the brunt of the environmental and social costs. The solution lies in decentralized, open-source alternatives that prioritize ethical design and circular economies, while creating neutral spaces for collaboration—echoing historical precedents like the International Space Station, where even Cold War rivals found common ground. Without such interventions, the AI chip wars will deepen global inequalities and accelerate the climate crisis, proving that the real 'chip' being carved is not silicon, but humanity itself.