While memory and optics companies have been largely acting together on the theme of artificial intelligence infrastructure in recent months, the picture started to change in the last week. While optical stocks such as Applied Optoelectronics (AAOI), Coherent (COHR) and Lumentum (LITE) gained upward momentum, memory companies such as Micron (MU) and Sandisk (SNDK) failed to make an equally strong breakthrough.
The first important element behind this divergence is the expectation that the USA may follow a more protectionist policy towards optical products originating from China. Regardless of whether the possible trading restrictions turn into an official ban, the message is important for hyperscaler companies. Large data center operators directing their optical component supply to US-based manufacturers could increase the order visibility and bargaining power of domestic manufacturers.
However, it may not be enough to explain the rise in optics with trade policy alone. The industry is still in the early stages of growth in the high-speed connectivity infrastructure needed in AI data centers. As the number of GPUs and the scale of data centers grow, the need for optical transceivers and connectivity solutions that enable data transfer between these systems is also increasing rapidly.
In this respect, the current position of the optical industry can be compared to the situation of memory about a year ago. As demand accelerates, manufacturers are trying to increase their capacity, but supply still lags behind demand. As AAOI management also emphasizes, there is not yet a sufficient supply of optical products to meet global demand. This creates a strong capacity increase and revenue growth potential for the sector for several years.
The story on the memory side has reached a different point. Strong demand for HBM and data center memory continues to support revenues, pricing power and margins from manufacturers like Micron. However, the important difference from the market's perspective is that a significant part of this growth is now reflected in financial results. Investors are therefore increasingly focusing on where the “next wave of growth” will come from.
This picture does not mean that the story of the rise in memory is over. The supply-demand balance may still remain in favor of producers and strong profitability may continue. However, on the optical side, the fact that the sector is at an earlier growth stage causes investors to price future capacity increase and demand potential more aggressively.
In short, the latest divergence in the market may indicate a capital rotation towards the new ring of the artificial intelligence infrastructure investment cycle, rather than an abandonment of memory. Following GPU and memory investments, the optical infrastructure that enables these systems to communicate with each other faster is increasingly becoming at the center of the artificial intelligence investment theme.
Currently, there are three issues with long-term potential that are not fully understood or priced: 1. Space, 2. Memory, 3. Optics. On the space side, obviously there will not be a boom as long as the pressure on the $SPCX side continues, but it is a sector that I think will be at a completely different level in the long term, especially as we move towards 2030. The second issue is the pressure on the memory side. The truth is that while the indices have been making new highs lately, the memory sector is not accompanying it. We see this very clearly in companies like $MU $SNDK falling between 30-50% from their peaks. Yes, in my opinion, this memory shortage, which is slightly present in current valuations, will continue for at least 1-2 years, and Micron's statement 1-2 hours ago confirms this.
Let's talk about the Photonics theme, which I follow closely. First of all, I can say that I think that a right choice in this sector will provide much better returns in the future. I'm even tracking tomorrow's $LITE balance sheet for my main portfolio.
He is my favorite player, but since the current multipliers are high for me, unfortunately I am waiting for the right levels to buy. There is also this reality: All of these companies are trying to increase capacity, which creates a need for capital. As far as I understand from the reports, there will be delays of 1-2 quarters on the CPO side. I want to buy exactly when this is being talked about. We also have US elections ahead of us, I think this will give us a much better entry level and I am creating my own road map accordingly. I think what you need to do is to make your own assessment and get involved at the level you determine. We have seen in the performances of Micron and Sandisk what can happen when supply is limited and demand is high. I think this situation also exists on the photonic side.
