So there's been a very obvious divergence recently. $NVIDIA(NVDA.US) has seen relatively smaller declines because it's the real 'shovel seller', and GPU demand is still there.

But the volatility is much higher for AI hardware chain companies like MRVL, COHR, GLW, and MU. While they are on the right track, their previous gains were too rapid, and the market needs to re-evaluate their valuations.

Take the optical communication sector, for example. Many think it's just hype, but the underlying logic hasn't changed. As AI data centers grow larger in the future, tens of thousands of GPUs will require high-speed connections. Copper wires will eventually be insufficient, making optical communication essential infrastructure. Therefore, the long-term thesis for companies like COHR, LITE, and GLW remains intact; it's just that their short-term stock prices are affected by sentiment.

The same goes for memory. The recent drop in $Micron Tech(MU.US) doesn't mean AI doesn't need memory. It's that cyclical stocks inherently have volatility, and the market trades ahead of supply-demand changes.

$Tesla(TSLA.US) is another example. Q2 deliveries exceeded expectations, but the stock didn't surge. The reason is simple: the market is no longer satisfied with just car sales numbers; it's waiting to see when future stories like FSD, Robotaxi, and Optimus will truly materialize.

So, in the current market, a drop doesn't automatically make something a buy, nor does a rise automatically make it expensive.

What you should really do is:

Identify the clear, long-term trends for the next few years, and then wait for good companies to be beaten down by sentiment.

The AI journey is still long, but the winners won't be all AI companies. It will be those with real technology, solid orders, and strong moats.

The biggest difference in investing isn't how much you make in a bull market, but whether you have the patience to wait for real opportunities when volatility hits.