Forget AMD And Micron — The Next Wave Of AI Winners Are The Ones Nobody Talks About – Advanced Micro Devi
However, that trade has become increasingly crowded. Valuations have expanded, money has flooded into cap-weighted ETFs, and the market has started treating “AI” and “tech” as interchangeable terms.
Yet the tech market is K-shaped. Just as the S&P 500 has split between AI winners and everyone else, the technology sector itself has broken into multiple independent cycles. Owning “tech” is no longer an investment thesis—it’s simply a basket containing businesses with entirely different economic drivers.
Solving the Three Bottlenecks
The irony is that this first wave of AI winners may also be the easiest to identify. The next phase won’t belong to the companies designing chips. It will belong to the firms solving the bottlenecks preventing those chips from operating at full scale.
According to the State of the AI economy report, AI demand is no longer speculative. Even after deducting the circular transactions, annualized GenAI revenue has surpassed $175 billion.
Meanwhile, hyperscalers have committed about $2 trillion of cumulative capital expenditure to build the infrastructure. In that compute supercycle, demand continues to outpace supply across multiple layers – yet bottlenecks are increasingly physical rather than computational.
The first is advanced packaging. Facing practical limits of single-die silicon, manufacturers are opting for stacking rather than a single enormous chip. This approach dramatically increases bandwidth while minimizing latency but requires sophisticated packaging technologies.
Without advanced packaging, next-generation GPUs simply cannot be manufactured at scale.
The second bottleneck is moving data. Training and inference clusters now contain tens of thousands of GPUs, and traditional copper networking is a constraint since power consumption and signal degradation sharply rise at higher speeds.
AI infrastructure is therefore migrating toward optical interconnects capable of handling 800G today and eventually 1.6-terabit connections. That trend creates an entirely different class of beneficiaries.
The final, least glamorous opportunity might be the most important one. As AI server racks grow, the heat they generate is increasingly inadequate for air cooling. Direct-to-chip cooling is thus transitioning from an optional upgrade to essential infrastructure.
Image via Shutterstock
