AI Technology Path Repeats Aluminum Metal History
AI is rapidly shifting from high-cost scarce resources to low-cost ubiquitous commodities, similar to the process of aluminum transitioning from a precious metal to a common material in the 19th century.
The inference costs of models like OpenAI have significantly decreased in recent years, with GPT-3.5 level performance dropping from several dollars per million tokens to just a few cents, making generative AI tools widely accessible to small and medium-sized enterprises and individuals.
Tech giants and startups are accelerating the deployment of open-source and efficient AI models, with funding shifting from early high-priced proprietary models to infrastructure optimization and application layer innovation. Traditional high-priced AI service providers are under pressure, while low-cost AI platforms and application developers benefit.
Source: Public Information
ABAB AI Insight
Charles Martin Hall and Paul Héroult invented the electrolytic process in 1886, causing aluminum prices to plummet from several dollars per pound to less than one dollar, transforming it from a royal luxury into an industrial staple. This technological breakthrough broke the extraction bottleneck.
Capital is being heavily invested in AI computing power, model compression, and edge computing, with companies reducing marginal costs through open-source models (like the Llama series) and hardware optimization, strategically shifting from "exclusive intelligence" to "scale-distributed intelligence" to seize dominance in the application ecosystem.
Similar to the journey of electricity or semiconductors from laboratory to commercialization, AI is currently in the early stages of transitioning from high-end laboratory applications to widespread infrastructure, with leaders focusing on energy efficiency and distribution rather than merely parameter scale.
Essentially, this represents a shift in technological substitution and pricing power: AI significantly lowers the threshold for "intelligent production" through algorithm and hardware iteration, releasing general computational intelligence like the Hall-Héroult process released aluminum, leading to a transfer of pricing power from a few laboratories to platform distributors and application integrators. This change is accelerated by the extension of Moore's Law and open-source competition.
ABAB News · Cognitive Law
Technological scarcity arises from extraction costs, not the resources themselves; breaking bottlenecks leads to abundance.
Yesterday's luxury often becomes tomorrow's public utility.
Winners do not sell scarcity but sell the structures that make scarcity cheap.