獵熊日記財經媒體分析者 shay 的趨勢領漲股的看法原文連結在跟貼有存下來僅作為立項開題

財經媒體分析師的觀點轉載節選，半信半疑地研究一下，原文連結在跟貼中：

Yesterday I did an interview with a journalist about what the next decade of the AI trade looks like.

Every market cycle starts with abundance and ends with constraint. The AI trade is no different where each leg reveals the next bottleneck:

• Compute (2023–2025)

This is the silicon era where $NVDA, $AMD, $TSM & $ASML are building the new industrial base of the digital economy. Compute is capacity so whoever controls it owns time.

• Power (2025–2027)

Then comes power where $VST, $CEG, $NRG & $NEE is about how it’s delivered, scaled and managed. This phase is about generation capacity & transmission lines.

• Cooling & Networking (2026–2028)

This phase is about keeping GPUs connected and efficient. $VRT, $AVGO, $ALAB, $MRVL & $CRDO are scaling the plumbing that moves data and manages heat. Optical interconnects, switches,m and liquid cooling systems all converge here where the next speed limit is heat and bandwidth.

• Data Movement (2028–2030)

After the pipes are built, attention shifts to how information flows where $SNOW, $PLTR, $MDB, $NET & $ORCL are building the operating rails for AI workflows, moving unstructured data across every enterprise and every cloud. Intelligence becomes a utility.

• Quantum & Energy Convergence (2030+)

Quantum expands the compute stack, sitting beside GPUs to solve the hardest optimization problems with $IONQ as my favorite name in the group. Nuclear gives the grid true baseload power by delivering the energy density needed to keep AI infrastructure running nonstop. So when you put it together its quantum breaks computational limits and nuclear removes power constraints.

But this last stage of the AI trade has been bid up too quickly. It’s the frothiest pocket of the AI theme right now that’s priced for perfection before the infrastructure even exists. The timeline is real but the market has sprinted years ahead of it.

Every leg of this cycle will have its winners but the edge comes from spotting the constraint before it becomes consensus.

Disclaimer: Community is offered by Moomoo Technologies Inc. and is for educational purposes only. Read more