The Transition from Pluggable to Co-Packaged Optics

↑
↑

  //science-technology.news-articles.net/content/2 .. sition-from-pluggable-to-co-packaged-optics.html
  Published in Science and Technology on Thursday, May 14th 2026 at 7:52 GMT by Seeking Alpha
      Locale: SWEDEN

The Transition from Pluggable to Co-Packaged Optics

For decades, data centers have relied on "pluggable" optical transceivers. These are modular components that plug into the faceplate of a switch, converting electrical signals to optical signals and back again. However, as speeds move toward 800G, 1.6T, and beyond, pluggable optics face a physics problem: power consumption and signal integrity. The distance the electrical signal must travel from the switch ASIC (Application-Specific Integrated Circuit) to the pluggable module results in significant energy loss and latency.

Co-Packaged Optics (CPO) solves this by moving the optical engine directly onto the same package as the ASIC. By reducing the distance between the compute engine and the optical interface, CPO dramatically lowers power consumption per bit and increases bandwidth density. For hyperscalers managing massive AI clusters, this efficiency is not merely a preference but a necessity for scaling.

Sivers Semiconductors and the Role of InP

Sivers Semiconductors operates in the specialized field of Indium Phosphide (InP) technology. While Silicon Photonics (SiPh) has garnered significant attention due to its compatibility with standard CMOS manufacturing, SiPh has a fundamental limitation: it cannot efficiently generate light on its own. It requires an external light source, typically a laser created from InP.

Sivers focuses on the development of Photonic Integrated Circuits (PICs) using InP. Their technology allows for the integration of lasers, modulators, and detectors onto a single chip. In the context of CPO, Sivers provides the "engine" that drives the optical transmission. Because InP offers superior performance in terms of light emission and modulation efficiency compared to silicon, it remains a critical component of the high-speed optical ecosystem.

Key Strategic Details

• InP Advantage: Indium Phosphide allows for the creation of active components (lasers) that Silicon Photonics cannot produce natively, making InP essential for the actual generation of light signals.
• CPO Adoption: The industry is moving toward CPO to reduce the "power tax" associated with moving data across a PCB from the chip to the front panel.
• Scalability: As a small-cap company, Sivers represents a high-leverage play on the adoption of CPO; if their components become integrated into mainstream AI server architectures, the volume growth could be exponential.
• Market Drivers: The proliferation of Large Language Models (LLMs) requires massive GPU clusters, which in turn drive the need for the ultra-high-speed, low-power interconnects that Sivers' technology enables.
• Architecture Shift: The move toward 1.6T and 3.2T speeds makes traditional pluggables increasingly impractical, forcing a pivot toward integrated optical solutions.

Market Implications and Risks

The opportunity for Sivers Semiconductors lies in the "hypergrowth" phase of AI infrastructure. As the industry transitions from the current generation of AI hardware to the next, the integration of CPO is expected to move from experimental phases to wide-scale deployment. This would transition Sivers from a niche technology provider to a critical supplier in the AI hardware stack.

However, the path is not without risk. The semiconductor industry is characterized by intense competition and rapid cycles of obsolescence. Sivers must navigate the transition from development to mass production while competing with larger established players who may develop alternative solutions or acquire competing technologies. Furthermore, the adoption rate of CPO depends on the willingness of hyperscalers to redesign their physical server architectures.

In summary, Sivers Semiconductors is positioned at the intersection of two massive trends: the expansion of AI compute and the necessity of optical interconnect efficiency. By providing the essential InP components required for Co-Packaged Optics, the company is targeting the very bottleneck that currently limits the scale of the AI revolution.