SemiAnalysis vs White Hair Stock God: What Real Gold Is Hidden in the CPO Report?

TL;DR

SemiAnalysis believes that the ramp-up of CPO will be slower than the market expects, and the transition window for NPO may be prolonged.
The white-haired stock god disagrees with this judgment, arguing that Nvidia and the supply chain are pushing CPO faster.
Related stocks: AAOI, LITE, COHR, GLW, MRVL, SIVE, AVGO, NVDA

The recent decline in optical stocks appears to be caused by a cold water splash on the CPO narrative, but in reality, the market is reassessing a more sensitive issue: whether the ramp-up in 2027-2028 is a performance realization period or merely a period of integration verification.

CPO (Co-Packaged Optics) itself has not been denied. The bandwidth, power consumption, and switching density pressures in AI data centers continue to increase, and the physical limits of copper cables and traditional pluggable optical modules have not disappeared. The issue is that the related stocks had previously risen with an implicit aggressive timeline: after Nvidia pushes CPO into the commercialization window, segments such as optical engines, lasers, silicon photonics, and switching chips are expected to ramp up to large-scale shipments quickly in 2027-2028.

The SemiAnalysis report on June 9 hit right at this pricing assumption. According to public reports, the study suggests that Nvidia's 800V DC and large-scale production of CPO may be delayed until around 2028-2029, while 400V DC is still on track for a ramp-up in 2026, with some NPO (Near-Package Optics) projects possibly accelerating. Following the report and the market volatility it caused, stocks like AAOI, LITE, COHR, GLW, MRVL saw high single-digit to double-digit corrections. The market is trading not on whether the "CPO direction is valid," but rather "how quickly can CPO turn into orders."

However, this is not a one-sided bearish outlook. The white-haired stock god and AI supply chain analyst Serenity (@aleabitoreddit) subsequently countered SemiAnalysis, arguing that it relies too heavily on conservative engineering models and underestimates Nvidia's ability to compress hardware cycles. Based on her interpretation of signals from Nvidia, Lumentum, Foxconn, and others, she emphasizes that CPO is still on its ramp trajectory in the second half of 2026, the second half of 2027, and 2028.

The value of this controversy is not to determine who "won," but rather to pull the valuation anchor of the optical chain back from an endgame narrative to a timeline validation: CPO will come, but the slope at which it arrives will determine the value distribution among NPO, pluggable modules, light sources, and switching chips.

Reassessment of the Timeline Behind the Decline of Optical Stocks

Over the past few months, the market's purchase of the optical chain is not driven by current revenue, but rather by the capital expenditure shift for the next generation of network architecture in AI data centers.

As model training and inference clusters expand, the communication pressure between GPUs, between racks, and within data centers continues to rise. The network is no longer just a supporting system outside of servers; it increasingly resembles an efficiency bottleneck in AI factories. The higher the bandwidth density and the lower the power consumption, the higher the limit for scaling power cluster units, which is also why CPO is being brought to the forefront.

The theoretical appeal of CPO is straightforward: bringing optical engines as close as possible to ASICs (Application-Specific Integrated Circuits), shortening the high-speed electrical signal path, and reducing the power consumption, loss, and signal integrity stress brought about by serial and parallel conversion circuits and copper traces. Compared to traditional pluggable modules, CPO has a better potential for power consumption and density in the high bandwidth era.

The market's challenge is that it easily trades "correct direction" for "certain ramp-up." Nvidia's official press release states that the Vera Rubin platform will introduce Spectrum-X Ethernet Photonics, and CPO switches have entered production for horizontal expansion and cross-cluster deployment in AI factories. A report from June 3 indicated that an Nvidia network business executive stated that Spectrum-X CPO switches had already been shipped to some partners, with production capacity expected to expand in the second half of 2026.

These signals are enough to prove that CPO is advancing, but they cannot be directly equated to the fact that production-grade large-scale orders are risk-free. For the capital market, there is a significant valuation difference between "entering production," "shipping to some partners," "customer evaluation," and "large-scale mass production." The pullback triggered by the SemiAnalysis report essentially indicates the market's beginning to differentiate these terms.

SemiAnalysis's Conservative Model: CPO Faces System Engineering Challenges

SemiAnalysis is not saying that CPO has no future. Its core judgment is more like: CPO's theoretical advantages are clear, but large-scale implementation will happen slower than the market imagines.

The reasons are not just that one or two devices are unprepared, but because CPO consolidates the complexity originally spread across modules, circuit boards, and systems into a more deeply coupled system. The higher the integration, the better the single-point performance, but the pressures on manufacturing, testing, maintenance, and supply chain flexibility will also increase.

The benefit of traditional pluggable modules is modularity. If a specific optical module fails, it can be replaced, and switching between suppliers is relatively easy. CPO is different; the optical engine is closer to the ASIC, and may even enter the same packaging system, with power consumption and density gains coming from this proximity, but the maintenance radius also expands. Once an optical component fails, it can affect not just an easily replaceable module but potentially involve more valuable switching chips and entire systems.

SemiAnalysis has previously emphasized serviceability, reliability, yield, and supply chain maturity in its CPO Book. Especially in the context of large-scale cloud vendors, performance is not the only metric. Major clients have high requirements for reliability and maintainability; if the failure rate, maintenance processes, and replacement costs in production environments are uncontrollable, no energy consumption model can delay its introduction.

InP lasers are also a point of contention. Laboratory-level port operating duration data can prove the technical feasibility, but does not equate to the ability to cover long-term operation, mass manufacturing, on-site maintenance, and supply chain redundancy in large-scale data centers. For investors, these differences are critical: laboratory validation proves the direction, while on-site reliability determines ramp-up.

Within the framework of SemiAnalysis, NPO and pluggable modules are not backward paths, but rather more realistic intermediate layers until engineering risks are fully released. CPO is theoretically superior, but if full implementation requires more time, the market must reprice these "less endgame, but easier to mass produce and maintain" solutions.

Serenity's Rebuttal: Nvidia Could Compress Hardware Cycles

Serenity's rebuttal does not deny that CPO has engineering challenges, but believes that SemiAnalysis underestimates Nvidia's organizational capacity in the AI hardware cycle.

Her logic is clear: Regular hardware introductions are indeed hampered by yield, reliability, and customer validation, but Nvidia is not an ordinary customer. It is both the definition maker of GPU cluster architecture and a core driver of network, switch, system integration, and supply chain dynamics. When the expansion of AI factories is blocked by network power consumption and bandwidth limits, Nvidia has enough economic motivation and voice in the industry to compress the traditional introduction cycle.

The evidence Serenity cites is twofold. The first layer consists of publicly cross-verified company statements, including Nvidia's official information about Spectrum-X Photonics entering production, as well as Lumentum's mention of CPO orders and delivery rhythm in Q2 FY26 information. Lumentum stated that it received CPO incremental orders worth hundreds of millions, with a delivery period in the first half of 2027, and the company's materials also indicated that CPO-related business is expected to enter a larger range of ramp-up in the second half of 2026.

The second layer involves interpretation of supply chain signals, such as the early delivery of optical switches from Foxconn to Nvidia. However, the specific scale of these signals, whether they are prototype samples or production-level orders, still needs more public information for confirmation.

This is also the key disagreement between Serenity and SemiAnalysis: SemiAnalysis believes that the variables of system engineering will naturally lengthen the cycle, while Serenity believes that Nvidia's supply chain execution capability will steepen this curve.

These two judgments do not completely conflict. Nvidia can bring CPO into production and customer validation earlier, and may also advance some horizontal expansion scenarios first, but this does not automatically mean that all AI data center networks will rapidly transition to CPO by 2027. The reliability tolerances and cost models for horizontal scaling, single rack expansion, internal and inter-rack conditions, and different customers vary, and the pace of introduction may also layer differently.

Serenity rebuts the overly conservative conclusion that "CPO will be significantly delayed," rather than proving that "CPO is already completely risk-free." For the market, this is enough to support the logic of a rebound from an oversold condition, but it is not sufficient to rewrite the aggressive revenue curve for 2027-2028 back into certainty.

Why NPO Suddenly Becomes Important

NPO has suddenly become important in this controversy because it just so happens to stand between the two logic sets of SemiAnalysis and Serenity.

It is neither the opposite of CPO, nor a simple continuation of traditional pluggable modules. The basic idea of NPO is to place the optical engine on a pluggable baseboard near the ASIC to shorten the electrical signal path, gaining a portion of power consumption and density benefits, while retaining better testability, replaceability, and supply chain flexibility.

If SemiAnalysis's conservative model is closer to reality, the deep packaging of CPO may slow down due to yield, maintenance, and reliability issues, making NPO a more realistic choice for a longer time. It allows ultra-large-scale cloud vendors to gradually accumulate experience in optical interconnection operations without overbearing all the risks of full CPO, and also provides existing optical module and optical engine suppliers with a longer window.

If Serenity's judgment regarding Nvidia's execution capability is more accurate, NPO may not disappear either. A more likely scenario is that NPO, CPO, pluggable, and copper interconnections coexist in different network layers. Nvidia's roadmap itself also shows that horizontal scaling of CPO can start first, while some scenarios of single rack expansion may still rely on copper cables or hybrid architecture in 2027-2028.

The impact on investors is that the pricing of the optical chain cannot simply be characterized as "CPO wins, others lose." Different technical routes correspond to different beneficiary links: CPO is more favorable to high-integration optical engines, laser light sources, silicon photonics, and the switching chip ecosystem. The extension of the NPO and pluggable windows may allow existing optical module vendors, connectors, materials, and some light source suppliers to continue enjoying orders and gross margin support.

The previous issue in the market was translating a technological endgame too early into a single performance slope. What has been reopened now is the valuation space for the intermediate route.

Production-Level Data is the Next Validation Point

This debate will not end in the short term merely relying on one report or a set of posts. SemiAnalysis reminds the market that the challenges of CPO lie in system engineering. Serenity reminds the market that Nvidia's supply chain organizational capacity may change the pace of traditional hardware introduction. The real divergence between the two must be verified through production-level data from the second half of 2026 to 2028.

The next crucial point is not "whether there are shipments," but rather the specifications of the shipments. Deliveries to some partners, client evaluations, initial production, scaling up, and large-scale deployment are entirely different stages. Nvidia's subsequent descriptions of mass production for Spectrum-X / Quantum-X Photonics, as well as the remarks from optical suppliers like Lumentum and Coherent regarding orders, production capacity, and gross margin, will be more important than single meeting wording.

It is also important to observe on-site reliability and maintenance data. If CPO's failure rate, replacement processes, yield curves, and total cost of ownership prove to be sufficiently stable in production environments, then the conservative model of SemiAnalysis will need to be revised. If this data lingers at the laboratory or small-scale validation level, the windows for NPO and pluggable modules will continue to be adjusted by the market.

The optical chain is now trading not on whether CPO will survive or not, but on the slope of the timeline. The next validation point will depend on whether "entering production" can evolve into sustainable ramp-up, and how the speed of these ramp-ups ultimately reflects in orders, gross margins, and client deployment specifications.

Although SemiAnalysis has raised concerns about CPO technology over the next two years, they still identified five semiconductor sub-sectors they remain optimistic about, namely:

Copper / AEC / ACC;

Pluggable optics / DSP;

CPO testing equipment;

Power gray space / UPS continuity;

Board-level VRM / Silicon-based power / Passive components

Specific stocks are integrated in the following image for readers' reference.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。