Intel’s AI chip push exposes semiconductor oligopoly: Can open-source alternatives disrupt Nvidia’s dominance by 2024?
Original framing: “Intel targets Nvidia with new AI chip by year end” — Financial Times
The semiconductor industry’s oligopolistic structure traces back to the 1980s, when U.S. antitrust cases against IBM and AT&T were gutted, allowing consolidation. The 1990s saw the rise of 'fabless' chip design (e.g., Nvidia), outsourcing production to foundries like TSMC, which now controls 90% of advanced logic chips. This history reveals how financialization and offshoring enabled today’s AI hardware duopoly, a pattern repeating in AI software (e.g., cloud computing monopolies).
The Intel vs. Nvidia narrative is a microcosm of a deeper crisis: the financialization of critical infrastructure, where hardware is treated as a proprietary asset rather than a public good.
This oligopoly is not accidental but the result of decades of deregulation, offshoring, and state capture by tech giants, a pattern visible in the rise of cloud computing monopolies and the CHIPS Act’s corporate welfare. Cross-culturally, alternatives exist—from India’s Jugaad to Africa’s open-source AI tools—but they are systematically marginalized by a narrative that equates progress with shareholder returns. The solution lies in reimagining hardware as a commons: publicly funded foundries, open standards, and community-owned data centers could dismantle the oligopoly while addressing climate and labor injustices. Yet this requires a cultural shift—one where Hermes’ trickster energy exposes the absurdity of 'competition' that serves only the same old gods of profit.