AI Chip Overinvestment May Be a Necessary Foundation, L&G CEO Says
Eric Adler argues that aggressive capital expenditure in semiconductors is not inherently wrong, despite growing bubble concerns.
The global race to build artificial intelligence infrastructure has sparked a fierce debate over whether the current spending spree is a strategic necessity or a speculative bubble. As cloud providers and enterprises pour billions into hardware, the industry is grappling with whether this capital expenditure can be justified by future returns.
Eric Adler, CEO of L&G Asset Management, stated in an interview with Maeil Business Newspaper that overinvesting in AI semiconductors is not inherently wrong. Adler offered this perspective while advising investors to maintain a "ballast"—a strategy of diversification and stability—to survive potential market volatility while continuing to pursue high-growth assets.
The Infrastructure Dilemma
The AI semiconductor market, currently dominated by NVIDIA, has experienced unprecedented growth as the world rushes to build the infrastructure required for Large Language Models (LLMs). This surge has created a divide between two schools of thought. Critics warn that the industry is inflating a bubble, arguing that if AI software revenue fails to scale in proportion to the massive hardware costs, a significant market correction is inevitable.
Conversely, proponents of the current spending trajectory argue that infrastructure must precede application breakthroughs. In this view, the hardware is the essential foundation upon which the next generation of productivity tools and AI services will be built. This suggests that what looks like overinvestment today is actually the prerequisite for tomorrow's utility.
Market Implications
The outcome of this investment cycle will likely determine the long-term viability of the AI economy. If the current trend leads to a glut of chips without a corresponding increase in demand for AI services, the industry could face a correction reminiscent of the dot-com bubble of the early 2000s. Such a scenario would see massive write-downs of hardware assets and a sharp contraction in semiconductor valuations.
However, if this infrastructure enables a new era of global productivity, the current spending will be viewed as a necessary foundation. In this scenario, the "overinvestment" phase serves as the catalyst for a broader economic shift, where the cost of the hardware is eventually dwarfed by the value of the software and services it enables.
Looking Ahead
Investors and analysts are now watching for signs of "AI monetization"—concrete evidence that enterprises are generating significant revenue from the AI tools they are deploying. While the hardware build-out continues at pace, the focus is shifting toward whether the software layer can catch up. Until a clear link between chip expenditure and software profit is established, the debate over whether the industry is overextended or simply preparing for a revolution will remain unresolved.