"CPO large-scale deployment has been postponed to 2027-2028"...

But 2028 was already the industry consensus, and that's the timeline sellers have been publicly stating all along. Jensen Huang has also been saying that copper connections won't be largely replaced by optical connections in the short term.

Now someone comes along and says "CPO still won't be ready by 2027," and everyone panics.

The key is, no one really thought mass production was possible by 2027 anyway. And even if the CPO story is disproven, shouldn't that be good news for pluggable optical modules?

The result is that optical modules, optical fibers, optical cables, even glass and bases are all falling together.

These small narratives in overseas markets are different from those in the A-share market. In A-shares, someone posts a screenshot and runs, making it hard to tell truth from falsehood, with no beginning or end.

Overseas, they are named and detailed, stated with absolute certainty, yet often accompanied by obvious logical errors.

For example, Citrini's "2028 Global Intelligence Crisis" that went viral in February this year—most friends should still remember it.

Google's TurboQuant paper from March 2026, which compressed the most memory-intensive part of AI inference to 1/6 of its original size, causing market panic that "HBM won't be needed as much," severely hit memory stocks.

It's all still vivid in our minds.