Gavin Baker, a managing partner of Altreides Management, said the artificial intelligence "bottleneck trade" has run its course [1].
This shift is significant because it signals a transition in how investors approach AI growth. For years, the most lucrative strategy involved betting on the companies that solved the physical and technical constraints of building massive data centers.
Baker, who was an early investor in SpaceX, said the bottleneck trade is a strategy focused on companies that alleviate supply-chain constraints for AI data-center builds [1]. These investments targeted the critical components, and services that previously acted as hurdles to rapid AI expansion [2].
Baker said the appeal of this specific strategy is now fading [1]. The primary driver for this change is that the supply-chain constraints that once fueled high demand for these specific services are easing [2]. As these bottlenecks resolve, the expected upside for the companies providing those solutions decreases [2].
Investors are now moving away from this approach as the market matures. The initial phase of AI development required a frantic rush to secure hardware and power, but that phase is evolving into a different stage of the economic cycle [3].
Baker said the strategy is winding down as the market adjusts to a new equilibrium [1]. While AI continues to grow, the specific financial advantage of betting on the "bottleneck" is no longer as potent as it was during the early build-out of the infrastructure [2].
“The AI 'bottleneck trade' has run its course”
The transition away from the bottleneck trade suggests that the AI industry is moving from an infrastructure-constrained phase to an application-driven phase. When the physical limits of data centers and chip supplies are no longer the primary obstacles, investor capital typically shifts from the 'picks and shovels' providers toward the companies that can generate a direct return on the AI software and services themselves.
