Citrini Research criticizes "bottleneck investors": The flow of AI value is becoming increasingly obscure, and blindly betting on computing power is extremely risky.

According to monitoring by Beating, Citrini Research—an independent research institution whose "AI Doomsday Report" released this February triggered fluctuations in the US stock market—publicly criticized the recent surge of "bottleneck investors" today: rushing to invest heavily without having any clear idea of what the industry will look like after computing power bottlenecks are solved is nothing but the blind leading the blind.

Known for its skill in extreme endgame projections, Citrini points out that the current technical roadmap of the AI industry is undergoing dramatic polarization. The market may either get stuck in a decade-long memory shortage, or, just as likely, face collapsing demand that forces foundational technological innovation to bypass existing hardware bottlenecks entirely. This extreme uncertainty is spreading from infrastructure to the basic structure of the entire market.

The biggest question is where massive profits will ultimately settle. At today’s table, large labs like Anthropic have every opportunity to dominate all links from fundamental infrastructure to applications. But just as reasonably, foundational models could become low-margin commodities while all core industry value shifts to application layer companies or even new innovators that have yet to emerge.

At the start of 2023, sidestepping endgame speculation and betting directly on infrastructure—the so-called "pick and shovel sellers"—was seen as the safest strategy. But Citrini stresses that instead of gaining clarity, the situation is only becoming foggier. Faced with a complex dynamic game, abandoning independent analysis of multiple possible endings and blindly following a linear "bottleneck" narrative brings enormous investment risks.