Cerebras’ IPO signals AI hardware consolidation amid extractive tech cycles: How venture capital and semiconductor monopolies shape AI infrastructure

Indigenous perspectives on AI hardware are largely absent in Western narratives, yet they offer critical insights into the environmental and social costs of semiconductor manufacturing. Communities in the Democratic Republic of Congo, which supplies 70% of the world’s cobalt, have documented the health impacts of mining on local populations and ecosystems, framing such extraction as a violation of territorial sovereignty. Traditional knowledge systems in these regions emphasize cyclical resource use and communal stewardship, starkly contrasting with the linear, extractive models driving AI hardware production.

The AI hardware boom mirrors historical patterns of tech cycles, where speculative capital inflates valuations before collapsing under structural inefficiencies. The 1980s semiconductor wars between the US and Japan, driven by state subsidies and corporate monopolies, resulted in overcapacity and financial losses, yet today’s AI hardware market repeats these dynamics with even greater stakes. The dot-com bubble of the late 1990s saw similar IPO frenzies, followed by a market correction that wiped out speculative ventures—raising questions about the sustainability of today’s AI hardware investments.

Cross-culturally, the framing of AI hardware as a purely economic phenomenon overlooks how different societies prioritize technological development. In China, AI hardware is tied to national security and industrial policy, with state-backed initiatives like the ‘National Integrated Circuit Industry Investment Fund’ driving investment. In contrast, European approaches emphasize ethical AI and open-source hardware, while African nations are exploring ‘leapfrogging’ strategies to bypass traditional tech monopolies through localized innovation hubs.

Scientifically, the focus on AI hardware IPOs distracts from the material and energy costs of training large models, which are often obscured by efficiency claims. Studies show that the carbon footprint of training a single large language model can exceed the lifetime emissions of multiple households, yet hardware innovations like Cerebras’ wafer-scale chips are marketed as ‘green’ without accounting for the full lifecycle impacts. The lack of standardized reporting on the environmental and social costs of AI hardware production remains a critical gap in scientific discourse.

Artistic and spiritual traditions offer metaphors for understanding AI hardware as a manifestation of human hubris and disconnection from ecological limits. Indigenous cosmologies, such as those of the Māori or Lakota, frame technology as a tool to be used in balance with nature, not as a means of domination. In contrast, Western techno-optimism often treats AI as a transcendent force, ignoring the material realities of its production and the spiritual void it may exacerbate in societies prioritizing consumption over community.

Future scenarios for AI hardware must account for the risks of monopolistic consolidation, which could stifle innovation and exacerbate geopolitical tensions. A decentralized model, where open-source hardware and community-owned data centers thrive, could mitigate these risks but requires policy interventions like antitrust enforcement and public investment in alternative infrastructures. The current trajectory risks a bifurcated AI ecosystem, where a few corporations control both the tools and the narratives of AI development, limiting societal agency.

Marginalised voices—such as factory workers in semiconductor plants, environmental justice advocates, and open-source developers—are systematically excluded from mainstream narratives about AI hardware. These groups highlight the human cost of AI production, from silicosis among chip workers in Taiwan to the displacement of communities near data centers in the US. Their absence in coverage reflects a broader failure to center the lived experiences of those most affected by the extractive logics of tech capitalism.