Optical communication concept stocks plummet, has the investment logic for AI infrastructure changed?

Optical Communications Sector Sees a Collective Pullback, and Market Sentiment Cools Rapidly

Recently, companies related to AI infrastructure have shown a clear pullback, with multiple optical communications, optical module, and high-speed interconnect supply-chain companies seeing their stock prices weaken in sync. Including Applied Optoelectronics, POET Technologies, Corning, Coherent, Lumentum, and Marvell Technology, among others, these firms have all recorded notable declines; for some individual stocks, the drop on a single day has exceeded 10%. This round of adjustment is not driven by deteriorating corporate earnings reports, but by a shift in the market’s expectations for the future development timeline of AI infrastructure, prompting capital to re-evaluate the valuations of related companies.

SemiAnalysis Report Prompts the Market to Reprice

(Source: SemiAnalysis_)

The core reason behind this market turbulence comes from a research report on AI compute infrastructure released by the research firm SemiAnalysis. The report notes that the CPO (Co-Packaged Optics, co-packaged optics) technology—highly anticipated by the market—may have mass-production progress far slower than originally expected.

The market had generally believed that CPO would have the opportunity to enter large-scale commercialization in 2027, but the latest research suggests that the time when it achieves truly comprehensive volume ramp-up may be pushed back to 2028–2029. This view directly challenges the market’s growth forecasts for the optical communications industry over the next several years, and has also become an important reason why optical-network-related companies were sold off.

Why Is the Development Speed of CPO Technology Being Questioned?

CPO is viewed as one of the key technologies for next-generation AI data centers. Its goal is to establish a more efficient method of data transmission between switching chips and optical components. However, SemiAnalysis believes there are currently three main challenges:

1. Yield issues in optical engine production— even in ideal conditions, it is still difficult to meet the demand from ultra-large-scale deployments.
2. The technical complexity of co-packaging switching ASICs with optical components is far higher than what the market originally estimated, making overall integration difficult.
3. In terms of cost effectiveness, CPO has not yet demonstrated enough advantages versus traditional pluggable optical modules, which may lead many companies to postpone their adoption timelines.

Therefore, the market has begun to reassess the growth pace of related companies over the coming years.

Potential Changes in AI Infrastructure Investment Directions\

In addition to CPO, the report also indicates that the previously highly watched 800V high-voltage DC power-supply architecture may be delayed in adoption. The research suggests that the more mature 400V DC power-supply solutions will still be the mainstream over the next few years, while alternative technologies such as Near-Package Optics (NPO) may have the opportunity to receive more capital expenditure support earlier. This means that AI infrastructure investment has not disappeared; rather, it may involve a shift in the technological roadmap. For investors, what needs attention in the future is not only whether the industry will grow, but also which technologies will ultimately achieve commercialization first.

Nvidia’s New Platform Progress Becomes a Key Focus for the Market

The report also mentions that the delivery schedule for Nvidia’s next-generation Rubin Ultra and Kyber platforms may be delayed. Since many optical communications, switch, and optical engine suppliers’ performance expectations are highly dependent on these new platforms, if delivery timelines are adjusted, the revenue recognition cadence across the related supply chain may also be affected. However, most market analysts believe that such changes mainly reflect adjustments in timing rather than the disappearance of AI demand. From a long-term perspective, demand for generative AI, intelligent agents, automation applications, and large-scale data center construction will continue to grow.

AI Investment Opportunities Are Expanding from a Single Technology to a Full Industry Chain

Over the past two years, market funds have flowed heavily into AI-related industries. As the industry gradually matures, the investment logic has also started to expand from a single popular technology to an entire ecosystem.

Besides the optical communications industry, investors are also paying attention to:

• AI chip design companies
• Cloud service providers
• Memory and storage device manufacturers
• Network switching and data center equipment firms
• Power and energy infrastructure companies
• AI software and application platforms

This trend of industry-chain expansion also makes investment opportunities more diversified.

Gate Stock Trading Officially Launches, Creating a New Global Tech Investment Entry Point

As AI and the technology sector continue to develop rapidly, an increasing number of digital-asset users are seeking more convenient ways to invest across markets. Recently, Gate officially launched stock trading services, further connecting multiple assets with global securities markets. With Gate stock trading, users can directly use USDT to invest in stocks and ETFs, without needing additional, cumbersome cross-border remittances or opening overseas brokerage accounts.

The platform currently supports more than 10,000 stocks and ETFs, covering major markets including Nasdaq, NYSE, NYSE Arca, NYSE American, and BATS.

Supports Fractional-Share Trading to Improve Capital Allocation Flexibility

In addition to a wide range of investment targets, Gate stock trading also supports fractional-share trading starting from as low as 0.01 share. For users who want to invest in high-priced technology companies but have limited capital, fractional shares allow them to gradually build positions and improve portfolio allocation efficiency. At the same time, diversified allocation across different targets can also help reduce the impact of volatility in a single industry on overall assets.

Summary

This pullback in the optical communications sector does not reflect an end to the AI boom. Instead, it reflects the market’s reassessment of technology implementation timelines and industry growth rhythms. From CPO and NPO to next-generation data center architectures, AI infrastructure remains one of the most important technology development directions for the coming years—though investment opportunities will gradually shift from single hot concepts to more comprehensive industry-chain layouts. With Gate stock trading officially launched, investors can participate directly in more than 10,000 stocks and ETFs using USDT, manage digital-asset and stock investments on a single platform, and more efficiently capture long-term investment opportunities driven by AI, technology, and global industry development.