BlockBeats news, on June 17, the hot trading around AI supply chain shortages may be drawing to a close, as some investors begin shifting focus from seeking the next shortage to identifying companies that will maintain long-term competitiveness beyond the AI infrastructure cycle.
Gavin Baker, Managing Partner at Altreides Management and an early investor in SpaceX, told TBPN that over the past year, the market has been focused on bottleneck assets in AI infrastructure, including DRAM, memory chips, and key material suppliers. However, he believes that these "AI bottleneck trades" are nearing their end.
Baker noted that Ajinomoto’s refusal to raise prices for a key chip packaging insulation material suggests that pricing power in some supply chain bottlenecks may be beginning to ease. The material, used to connect processor and chip packaging layers, had drawn investor attention due to surging demand for AI chips.
He said that in the past, the market's game was about finding the "next bottleneck," but the more important question for the next phase is which companies will retain lasting franchise value after these bottlenecks disappear.
This perspective has cooled the recent surge in storage and AI materials stocks. Storage stocks such as Micron and SanDisk have risen sharply this year, driven by a reassessment of supply and demand due to capital expenditures on AI data centers, demand for HBM, and long-term procurement agreements. However, as stock prices have risen rapidly, the market is now beginning to discuss whether the trade has become overly crowded.
Baker also noted that the next major focus in AI infrastructure may shift toward the practical ability to deliver computing power and electricity. He highlighted SpaceX’s potential in ground- and orbit-based AI data centers, stating that the market will closely monitor its speed in scaling up gigawatt-level computing capacity.
This means AI trading may shift from simply betting on shortages to competing more broadly on infrastructure: those who can secure land, electricity, GPUs, and data center capacity faster may gain a more advantageous position in the next phase of AI investment.