The US stock market has pulled back for two days, and many high-quality stocks are once again presenting opportunities.

Recent trading characteristics are very clear:

Strong sectors still see capital inflows after a pullback, while weak sectors tend to follow the market down once it weakens.

The current market is dominated by group-based trading.

AI hardware, semiconductors, servers, and data center energy chains are the main capital themes.

Software, streaming media, some internet and traditional growth stocks lack incremental narratives in the short term.

Even if weak sectors experience a brief rebound, they tend to fall back once the market turns weaker.

The AI rally is not over, but divergence will intensify.

Semiconductors, AI servers, data centers, power, cooling, and optical modules remain the most important market themes.

But capital will no longer chase indiscriminately,

Going forward, the market will pay more attention to who is truly “selling shovels,”

Who is just storytelling,

Who can turn AI demand into cash flow and profits,

Who is merely spending money to buy computing power.

The biggest current market problem is not a sudden deterioration of fundamentals, but that the recent two months of gains have been too smooth.

Indices have risen consecutively, volatility has been suppressed, investors’ defensive willingness has declined, and positions are gradually becoming crowded.

In this environment, any macro variables, earnings report changes, or shifts in options structures

could trigger short-term adjustments.

Therefore, a market correction this week is healthier for future gains,

A decline is an opportunity to get on board, so the focus remains on which stocks to buy.

Previously, the market was trading: as long as it’s related to AI, valuation could expand.

Now, the market is trading: AI-related growth is not enough; stronger growth, higher guidance, and better profit margins are required.

In the first half of AI, buy GPUs; in the mid-term, buy networks; in the second half, look at optical interconnects.

CPO is not benefiting all companies with “optical” in their names,

But those caught in the exchange chips, DSPs, silicon photonics, lasers, and high-yield manufacturing segments are the first to be revalued.

That’s why I keep saying buy Broadcom; Broadcom is no longer just an ordinary chip stock,

But a key leader in AI, ASICs, AI network chips, and customized accelerators.

Marvell follows the same logic, but this stock has been overly hyped,

In reality, there isn’t that much room #我的Gate交易时刻 for growth.