The recent decline in optical communication resembles a technical bear market, not a fundamental one.

I believe the correction in optical stocks is nearing its end.
Take $LITE (Figure 1) as an example — it has been consolidating for 58 days since the correction started on May 12, with a share price pullback of about 37%. Late July to mid-August is the earnings window for optical companies. A pre-earnings expectation kill and valuation kill is actually a healthy thing, releasing short-term overheated sentiment and laying the foundation for the next rally.
I think we are very close to the bottom, but I don't rule out a second leg down. Technically, the first target above is still the daily EMA20 and EMA50 (Figure 3). If it fails to break through on the first attempt and meets resistance there, a retest and confirmation is also a normal move.

Now look at $NOK. It has been falling for six consecutive weeks, but there has been a change in these weeks: trading volume is gradually shrinking, indicating that selling pressure is slowly weakening. With earnings still about two weeks away, a technical rebound driven by short covering is not surprising. The stock price is now starting to attempt a breakout of the short-term downtrend line. Next, focus on resistance near the EMA20 (approximately $13.2).

Yesterday I cited the NOK 1998 example in the comments (Figure 2). Let me explain again today.
From late 1997 to mid-1998, NOK first experienced a rapid rally, then entered a consolidation of about 10 weeks, with a maximum drawdown of about 36%. Now NOK's pullback from the high has also reached a similar level, about 34%. You all know the story after the consolidation — it rallied from October 1998 all the way to June 2000. Of course, I’m not saying history will repeat exactly, let alone that the same gains can be replicated. I just want to illustrate that a 30%+ correction during a bull market does not mean the long-term trend is over.

Finally, let’s talk about $MRVL (Figure 4).
I believe it is still in a continuation phase of an uptrend, more like forming a large Bull Flag. The previous gap has been largely filled. It will likely continue to oscillate and consolidate in this range for a while, waiting for new catalysts.
If you still believe in the long-term logic of AI infrastructure, you need to watch corporate earnings, big tech capex, and orders — not daily price swings.
Starting at the end of this month, Microsoft, Meta, Amazon and other giants will report earnings one after another. I still tend to think the AI arms race won't end so quickly. Two pieces of evidence I posted yesterday: Meta plans to invest $10 billion to build a data center in Canada, and Amazon plans to invest $5 billion to build a data center in the Philippines. If computing power is truly oversupplied and capex is being cut as Wall Street says, why are the giants still increasing capital expenditure? Use logic to think. Whoever significantly cuts capex first will likely see the gap in models, computing power, and ecosystems widen further in the coming years. So I’m more focused on management’s forward guidance on CapEx.

Summary
Remember, in a bull market, corrections and washouts are for better gains. As long as you believe it’s still a bull market and hasn’t turned bearish, you should not be pessimistic about any correction. Take this Bitcoin bull cycle as an example: in 2023, it fell from 32,000 to 25,000; in 2024, during the rally, it dropped from 74,000 to 49,000; in 2025, from 108,000 to 75,000. Everyone screamed the bull market was over, but each time it turned out to be just a consolidation mid-way. Most people know the four-year cycle is supposed to end by late 2025, so isn’t it simple to be a patient holder with spot until the cycle ends? Yet 90% can’t do it because they are heavily swayed by the pessimistic sentiment in between. Now it’s the same with AI, except the anchor has shifted from the four-year cycle to whether big tech capex has been significantly cut, whether AI commercialization revenue and token growth have slowed, whether HBM and data center demand has weakened, whether AI infrastructure companies’ profits are still growing, and whether global liquidity is still improving (btw, M2 money supply posted its largest monthly increase in five years). None of these have changed, so why would you think the AI bull market is over? I see it as a healthy valuation-adjustment correction.
If the bull market isn’t over, then wait for the structure to complete. Don’t be scared away by the washout. The real main uptrend always begins after a period of consolidation that tests people’s patience, and then rises amid doubt.