The "biggest rival" of optical modules—when will CPO truly achieve mass production?

```

Co-Packaged Optics (CPO) technology is seen as a potential disruptor to traditional optical modules, but its path to large-scale mass production is much longer than the market anticipated.

According to Digitimes, driven by the wave of upgrades in cloud AI infrastructure, the commercialization process of CPO technology continues to attract close attention from the market. NVIDIA is currently the most proactive company advancing CPO mass production, continually pressuring suppliers to have mass production capabilities starting in 2025.

However, according to supply chain sources, the actual mass production scale of CPO-related products will remain very limited in 2026, and full-scale rollout will still take time.

The core bottleneck restricting CPO mass production is yield. For CPO to achieve large-scale deployment in AI data centers, production yield must reach a sufficiently high level for its cost-effectiveness to surpass existing solutions. This issue has now been recognized by the entire industry chain as the biggest obstacle.

Yield bottleneck restricts scale, mass production progresses slower than expected

Despite cloud AI vendors looking to introduce silicon photonics (SiPh) technology to simultaneously improve cost efficiency and the upper limit of computing performance, the gap between supply chain reality and market expectations remains significant.

From a supply chain perspective, the actual mass production scale of CPO-related products in 2026 is extremely limited, and widespread rollout is still a distance away. The high difficulty in production and verification stages means there is significant room for yield improvement, and only when yields reach a sufficiently high level can CPO's cost-effectiveness truly surpass existing solutions. This yield issue is currently widely recognized by all parties in the industry as the biggest bottleneck for large-scale implementation.

CPO supply chain participants pointed out that were it not for the urgent demand driven by cloud AI upgrades leading to large-scale investments in CPO technology, the timeline for related technological breakthroughs would likely have been substantially delayed. From this perspective, NVIDIA's proactive push has had a significantly positive effect on the entire ecosystem.

NVIDIA is the most aggressive, Broadcom and Marvell are relatively conservative

Among major chip manufacturers, NVIDIA is the most assertive in advancing CPO deployment. Starting from 2025, NVIDIA continues to urge suppliers to prepare mass production capabilities and emphasizes to customers the significant computing power advantages that CPO brings to its system architecture, putting considerable pressure on suppliers. Supply chain sources predict that NVIDIA's mass production advancement will outpace other manufacturers.

Broadcom has been deeply involved in the CPO field for many years and has now begun preliminary shipments to customers. Marvell Technology and MediaTek are also making moves in this technology. However, according to supply chain sources, Broadcom and Marvell are relatively less aggressive in promoting their respective CPO solutions.

CPO-related manufacturers emphasize that regardless of which company is the first to break through, market demand will continue to grow, and chip manufacturers and ecosystem partners will all benefit. At this stage, the only factor limiting industry growth remains the supply-side ceiling of capacity and yield.

In addition, from the demand side, the continued expansion of cloud AI infrastructure provides a clear and strong market driver for CPO technology. The silicon photonics ecosystem is eager to see substantive revenue contributions, and cloud AI vendors are likewise motivated to break through current computing power and cost constraints by adopting SiPh technology.

However, the key variable determining whether CPO can truly become the "terminator" of optical modules has shifted from the demand side to the supply side. The speed of capacity ramp-up and yield improvement will directly determine the timeline for CPO technology to move from the laboratory to large-scale deployment in data centers.

Risk Warning and DisclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice and does not take into account the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular circumstances. Investing based on this article is at your own risk. ```