CPO Long-term Logic Unchanged, but Short-term Shipment Expectations Need Significant Downward Revision

Overview

This research report responds to recent market concerns regarding the delayed introduction of Co-Packaged Optics (CPO) technology. Morgan Stanley believes that although the long-term growth logic for CPO remains solid, investors' overly high shipment expectations in the short term need to be reset. Through quantitative analysis of capacity and yield bottlenecks, the report points out that actual shipments in 2027 will be significantly lower than market consensus, and the true industry explosion period will be postponed until after 2028. Based on this judgment, the institution recommends distinguishing between short-term sentiment fluctuations and long-term industry trends, and continuing to hold supply chain targets with core positioning advantages.

Core views

Huge gap in shipment expectations: The report's model calculations show that global optical engine (OE) shipments in 2027 (including Scale-up and Scale-out solutions) are expected to be only 6 million to 7 million units. In contrast, current general investor expectations are as high as 20 million to 30 million units. This huge expectation gap means CPO-related stocks may face pressure from sentiment reset and stock price adjustment in the short term. Manufacturing yield is the core bottleneck: Although TSMC plans to expand Photonics Integrated Circuit (PIC) capacity to 10,000 wafers per month in Q1 2027, this does not equate to effective output. The report emphasizes that the yield of the SoIC (System Integrated Chip) packaging stage is currently only between 50%-60%, and the yield of the downstream assembly stage is even lower at 20%-50%. The yield levels of these two stages are the key swing factors determining final CPO shipments, and are also the main reason why the volume release pace is slower than expected. Technology route is in a transition coexistence period: The report believes that 2026 to 2028 will be a transition stage where multiple interconnect solutions coexist. During this period, traditional pluggable optical transceivers, various optical interconnects (CPO/NPO), and copper cable connections will jointly serve market demand, as mainstream bandwidth solutions remain at the 1.6T/3.2T level. CPO technology still needs time to complete the ramp-up, and its true scale explosion node is expected to appear starting from 2028. Differentiated stock views: Despite short-term disturbances, the report still maintains an Overweight rating on CPO core enablers, including TSMC (2330.TW), ASE (3711.TW), Accelink Technologies (3363.TWO), Wontech (6187.TWO), Innodisk (6223.TWO), FIT Hon Hai (7769.TW), and Innodisk (6515.TW). It specifically mentions that Accelink Technologies' CPO revenue is expected to start ramping up at the end of Q3, and performance before that may be relatively flat. For A-share Suzhou Tfc Optical Communication (300394.SZ), while recognizing its opportunities in the CPO component field, considering its exposure in the pluggable optical module component business and the significant valuation premium relative to A-share peers, it is given an Equal-weight rating.

Analysis framework

The report adopts a typical 'Capacity-Yield-Effective Output' transmission analysis framework. Unlike the market's simple linear extrapolation of capacity expansion logic, the institution sinks the analysis focus to the 'friction coefficient' of the manufacturing stage. Specifically, analysts did not just look at TSMC PIC wafer input volume, but used it as a starting point, sequentially multiplying by the two discount factors of SoIC packaging yield and downstream assembly yield, thereby deriving effective shipments closer to real delivery capacity. This method reveals the huge gap between nominal capacity and actual supply, explaining why terminal volume release remains slow even if the upstream actively expands capacity. At the same time, combined with judgments on the evolution pace of mainstream bandwidth specifications (1.6T/3.2T), the time window for new technology penetration is determined, thereby calibrating market over-optimism.

Methodology notes

• Industry/Industrial Analysis FrameworkSupply-demand framework

  Effective Output = Nominal Capacity × Comprehensive Yield

  In emerging manufacturing fields such as semiconductor advanced packaging, equipment or wafer capacity cannot be directly equated to market supply. The report corrects linear predictions based on nominal capacity by introducing low yields in SoIC and assembly stages as key loss factors, which is a core method for understanding early supply bottlenecks in high technical barrier industries.

• Industry/Industrial Analysis FrameworkPenetration S-curve

  Coexistence Period of Old and New Technologies and Explosion Inflection Point

  The replacement of old technology by new technology is not instantaneous; there is often a long 'coexistence period' in between. The report points out that 2026-2028 is a transition phase where pluggable modules and CPO coexist, and CPO will not enter the accelerated climbing phase of the S-curve until after 2028. This reminds investors to distinguish the time lag between 'technical feasibility' and 'scale commercialization' when tracking industry trends.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• TSMC (2330.TW)

  CPO core enabler, providing PIC wafer manufacturing and SoIC packaging capabilities

  Strengths

  Possesses PIC capacity expansion plans and advanced packaging technology dominance

  Weaknesses

  SoIC yield is still in the 50-60% ramp-up period, limiting short-term effective output

  Risks

  Yield improvement speed is slower than expected

• Accelink Technologies (3363.TWO)

  Key supplier of CPO optical engines and fiber optic connectors

  Strengths

  CPO revenue expected to start ramping up at the end of Q3 2026

  Weaknesses

  Revenue performance is flat before mass production, lacking short-term explosive power

  Risks

  Mass production progress delay

• Suzhou Tfc Optical Communication (300394.SZ)

  Potential key participant in CPO components

  Strengths

  Has good entry opportunities in the CPO component field

  Weaknesses

  Pluggable optical module component business still accounts for a high proportion, and valuation has a significant premium compared to A-share peers

  Comparison

  Valuation significantly higher than A-share industry average

  Risks

  High valuation digestion pressure, traditional business drag

• ASE Technology Holding (3711.TW)

  Core manufacturer in CPO packaging and testing stages

  Strengths

  Advanced packaging leader, benefiting from long-term CPO trends

  Risks

  Downstream assembly yield bottlenecks affect order delivery

Key data

• 2027 Optical Engine Shipment Forecast6-7 million unitsIncludes Scale-up and Scale-out solutions, far below market expectations of 20-30 million units
• TSMC PIC Capacity Plan10k wpm (10kwpm)Expected to be achieved in Q1 2027
• SoIC Packaging Yield50%-60%One of the key swing factors constraining effective output
• Downstream Assembly Yield20%-50%Another major manufacturing bottleneck constraining final shipments
• Mainstream Interconnect Bandwidth Specifications1.6T / 3.2TMainstream solutions during the 2026-2028 transition period, limiting CPO's short-term substitution speed

Impact & implications

The report believes that the CPO sector will undergo a necessary 'expectation calibration' process in the short term. Due to the actual shipment pace being far slower than the previous optimistic pricing by the capital market, related stocks may face pressure from weakening sentiment and valuation correction, especially before the first half of 2027, where performance realization of supply chain companies may be low. However, in the long run, this adjustment does not change the path of CPO as the core technology for next-generation high-speed interconnects. The logic of explosive growth after 2028 still holds. For investors, this means the current volatility is more of a timing mismatch rather than a falsification of industry direction. Leading enterprises with core technical positioning (such as advanced packaging, testing, optical components) will still benefit from long-term structural growth after passing through the yield ramp-up and expectation reset period.

Risks

• SoIC packaging yield and downstream assembly yield improvement speed is slower than expected, leading to further downward revision of 2027 shipments
• CPO technology introduction delay, leading to continued market sentiment depression and related stock selling
• 1.6T/3.2T pluggable solution lifecycle extension, squeezing CPO's market share during the transition period

What to watch

• Improvement progress of TSMC SoIC packaging yield and downstream assembly yield
• Actual ramp-up situation of Accelink Technologies CPO revenue at the end of Q3 2026
• Whether CPO enters the scale explosion phase as expected starting from 2028