Analyzing the Risks of AI Sector Consolidation and Infrastructure Overcapacity An exploration of the symbiotic relationships between AI giants like Nvidia and Corning, the potential for a market bubble reminiscent of the 1990s, and the debate over capital expenditure sustainability. The current landscape of artificial intelligence is witnessing a trend of corporate interconnectedness that is becoming increasingly unsettling for market observers. A primary example of this is the deepening partnership between Corning and Nvidia. Corning has announced a massive ten-fold increase in its manufacturing capacity for fibre optic cable equipment, with new facilities slated for North Carolina and Texas. This expansion is particularly noteworthy because Corning had previously indicated it would only increase production if it received prepayments from its clients.Analysis from Citi suggests that Nvidia is essentially funding this growth, evidenced by the fact that Nvidia has the option to purchase 15 million Corning shares at 180 dollars each, and that the new equipment is specifically designed for Nvidia chips. This arrangement mirrors the 1990s in conflicting ways. On one hand, it resembles Intel's strategy of risking everything to dominate production capacity, which granted them immense pricing power.On the other hand, it evokes the telecom crash of the 90s, where vendor-financed capacity far exceeded actual demand, leading to a catastrophic collapse. Corning now faces the risk of creating a surplus of infrastructure that the market cannot sustain over the long term. Beyond the Corning and Nvidia deal, there is a broader and more systemic concern regarding the cozy arrangements among the dominant players in the AI space.The current structure resembles the Japanese keiretsu of the 1980s, characterized by dense networks of companies holding shares in one another. Historically, such arrangements in Japan and South Korea led to economic paralysis, as they protected poorly performing companies and made corporate reform nearly impossible, eventually resulting in massive overcapacity and decades of stagnant market returns.We see similar patterns today with Microsoft's long-term tie to OpenAI, the billions invested by Amazon and Google into Anthropic, and the overlapping interests of Oracle, Nvidia, and OpenAI. When a small group of powerful entities attempts to lead an economic revolution while simultaneously limiting competition through these mutually supportive structures, it creates a fragile environment. While these hyperscalers will continue to purchase equipment in the medium term, the long-term stability of this closed ecosystem is questionable.Another critical point of contention is the tension between massive capital expenditure and actual earnings growth. Lisa Shalett, the chief investment officer at Morgan Stanley Wealth Management, has expressed a bearish outlook, arguing that many investors are ignoring macro variables such as inflation, oil prices, and Federal Reserve policy. She posits that capital expenditure booms are inherently self-undermining because spending becomes more volatile as the trend matures.Moreover, the transition of software giants like Meta and Microsoft from capital-light business models to asset-heavy enterprises—due to the immense cost of building and maintaining data centres—could lead to lower stock multiples in the future. Conversely, some analysts, like Hugo Ste-Marie from Scotiabank, argue that the current record stock prices are justified by strong profit momentum, noting that 84 percent of S&P 500 companies are reporting earnings above consensus estimates.This suggests a divide between those who see current growth as a sustainable new era and those who see it as a cyclical peak. On a separate note, the discourse surrounding digital distractions in education provides an interesting counterpoint to these corporate narratives. Despite the well-documented destructive nature of social media—including the spread of misinformation and the erosion of public discourse—recent experiments with cellphone bans in primary and high schools have yielded surprising results.Contrary to the expectation that removing phones would boost academic performance, these experiments have shown no verifiable improvement in educational gains. This suggests that the relationship between technology use and learning is more complex than a simple cause-and-effect correlation, mirroring the complexity and unpredictability currently seen in the AI-driven financial markets