#英伟达市值重返5.4万亿美元 Nvidia's market capitalization has returned to $5.4 trillion, marking the ultimate validation of the industrial logic and capital pricing of the AI computing power sector. The rally in the AI computing power sector has not yet peaked, but it has shifted from the “broad-based gains” stage into a stage of “differentiation” and “volatility.” Overall, the market will exhibit a “wave-like” progression, with the medium- and long-term outlook remaining promising, while facing pressure from valuation digestion and sentiment volatility in the short term.
I. The underlying logic behind how long the rally can continue
1. Industrial cycle support (long-term logic)
AI computing power demand is far from peaking. It is extending from “large model training” to “model inference” and “AI agents,” with Token consumption growing exponentially and continuously driving demand for computing power. The continued expansion of capital expenditures by global technology giants (cloud providers) provides long-term demand support for the upstream computing power industry, and the industrial cycle logic of the computing power sector remains intact.
2. Earnings realization and valuation dynamics (medium-term logic)
The current valuation of the computing power sector has fully priced in growth expectations for the next two to three years. The sustainability of the short-term rally depends heavily on continued earnings realization (such as actual orders, gross margins, and cash flow across various segments) as well as changes in macro liquidity. As the base becomes higher, the growth rate of the computing power sector may slow marginally, and the market will become more selective, with the rally exhibiting a volatile upward trend characterized by “two steps forward and one step back.”
II. Key indicators for monitoring the market's evolution
1. Sustainability of capital expenditures (CapEx): Monitor the implementation of capital expenditures by overseas cloud providers (Microsoft, Google, Amazon, Meta, etc.) and domestic internet giants, as this is the most direct leading indicator of computing power demand.
2. Earnings realization: Monitor the order visibility, actual revenue, and profit realization across segments of the computing power industry chain (such as optical modules, memory chips, advanced packaging, and liquid cooling). Earnings validation will determine internal differentiation within the sector.
3. Macro liquidity: The AI computing power sector is a high-valuation growth sector and is relatively sensitive to U.S. Treasury yields and expectations for Federal Reserve rate cuts. Changes in liquidity will directly affect the room for valuation recovery in the sector.
III. Structural differentiation within the sector
1. Segments with strong growth and high earnings visibility (likely to advance further): Upstream core hardware (such as high-end optical modules, HBM high-bandwidth memory, and advanced packaging) and computing power leaders with core technological barriers have high earnings realization due to tight supply and demand and pricing power advantages, giving them strong downside resilience and long-term allocation value.
2. Purely conceptual segments or those with weak earnings realization (higher risk): Some stocks driven solely by AI concept speculation and lacking actual orders and earnings support are prone to significant pullbacks after valuation recovery as capital flows recede and sentiment declines.
The long-term potential of the AI computing power sector remains broad, but it will be difficult to replicate the one-way surge of the past in the short term. Investors should lower their short-term return expectations, capture the convergence of industry trends and earnings realization, and avoid blindly chasing highs. $NVDA ‌