Citi recorded key takeaways from its interview with Lightmatter's management. Management stated that Near-Packaged Optics (NPO) will enter the market in 2027 and achieve widespread adoption in 2028, while the timeline for Co-Packaged Optics (CPO) scale-up remains unchanged at 2028–2029. Management characterized lasers and optical fibers as the "next HBM," with expectations of tight supply supporting a scarcity premium. The bottleneck for CPO lies in testing capacity rather than the technology itself, presenting an underappreciated investment opportunity in related test equipment.
Citi believes that NPO has replaced CPO as the preferred near-term trading thesis, while lasers and optical fibers are being positioned as the "next HBM," with expectations of supply shortages driving a new valuation narrative for related assets.
On September 8, Citi’s Atif Malik team published a research report summarizing key points from interviews with Lightmatter CEO Nicholas Harris and CFO Simona Jankowski during the Citi Global TMT Conference.
Management at photonics interconnect startup Lightmatter indicated that Near-Packaged Optics (NPO) is expected to enter the market in 2027 and achieve large-scale adoption in 2028. Meanwhile, the window for scaling Co-Packaged Optics (CPO) remains set for 2028–2029.
Management also characterized lasers and optical fibers as the "next HBM," implying that tight supply conditions will support a scarcity premium for these assets.
NPO replaces CPO as the mainstream near-term trading direction
According to Citi Research, prior to the 2026 Optical Fiber Communication Conference (OFC), the market generally expected CPO to lead scale-up efforts between 2027 and early 2028. However, following the conference, market focus has clearly shifted toward NPO.
Lightmatter management stated that NPO platforms will begin to emerge in 2027, followed by high-volume, widespread deployment in 2028. The timeline for CPO application in scale-up scenarios remains within the 2028–2029 range.
The implementation of OCI industry standards is considered a significant catalyst for accelerating current design activities, as previous uncertainties at the architecture level had suppressed customer investment.
Lightmatter anchors its economic rationale primarily on the cost savings delivered by Bidirectional (BiDi) fiber technology. Management aims to achieve a substantial reduction in unit cost per gigabit per second (Gbps) compared to existing pluggable transceivers (benchmarked at approximately $0.5/Gbps).
Its Passage L20 NPO engine provides 6.4 Tbps of bandwidth and utilizes BiDi technology, enabling simultaneous transmission and reception over a single fiber optic cable, thereby eliminating the need for separate transmit and receive fibers.
Management stated that this approach halves fiber optic usage. Based on the overall scale of data centers, management quantified this efficiency gain as approximately $1.6 billion in network cost savings per gigawatt.
In this context, management characterized lasers and fiber optics as the "next HBM," with their scarce supply attributes drawing market attention.
The bottleneck for CPO is test infrastructure.
While the prevailing external interpretation of slow CPO progress often focuses on technological immaturity, management explicitly pointed out that the core constraint on CPO scalability is currently testing capacity, not the technology itself.
Current annual production of CPO switches ranges from only a few thousand to approximately 10,000 units, whereas management estimates that industry-required capacity is about 100 times the current level. Suppliers of wafer-level test equipment remain severely scarce. This assessment suggests that investment opportunities in the test equipment segment may be undervalued by the market.
At the industry standards level, Lightmatter participates in the Open Compute Project (OCP) Silicon Photonics Initiative, which aims to unify system-level design specifications and avoid redundant investments across the industry in system and rack-level engineering.
Management introduced that this collaboration has expanded from an initial nine partners to approximately 19, and a joint specification document spanning 300 pages has been released, positioned similarly to NVIDIA's reference design manuals.
Meanwhile, Lightmatter entered into a partnership with NVIDIA regarding NVLink Fusion in June 2026.
Management stated that NVLink Fusion was previously limited by copper cables, allowing only short-distance transmission within racks. The introduction of fiber optic interconnects will break this physical boundary, extending the scaling domain beyond the reach of single-rack copper cabling. Management cited over one million hours of error-free data transmission as evidence of technical readiness.
Strategic shift in laser technology toward integrated "ultra-large-scale integration" photonic devices
Regarding the laser technology roadmap, management assesses that the optical power of single-device lasers has approached its physical limit—approximately 400 mW, with a threshold around 500 mW. This implies that further bandwidth enhancements will require an increase in the number of lasers.
Lightmatter’s solution involves laser arrays based on GaAs materials using 300mm CMOS wafers, integrating 128 lasers per chip. Its latest products achieve a bandwidth of 16 Tbps per module and are under development for various packaging formats, including OSFP.
Management positions the photonic interposer as a unique technological product in the current market, claiming it eliminates the shoreline constraints on laser placement and offers approximately ten times the bandwidth density of alternative solutions. They state that no other company currently offers comparable technology.
As design activities for NPO (Near-Package Optics) and CPO (Co-Packaged Optics) intensify, customer demand pull for interposer designs is increasing concurrently.
At the modulator architecture level, the OCI MSA has formally established the microring modulator as the industry-default architecture, prioritizing it over electro-absorption modulators.
Management stated that Lightmatter’s thermal control system can maintain the stability of microring modulators under extreme gradients of 2,000 degrees Celsius per second, even in scenarios involving stacked packaging with high-power advanced-node accelerators.
Looking to select stocks or analyze your holdings? Want to understand the opportunities and risks in your portfolio? For any investment-related questions,just ask Futubull AI!
Editor/lambor