qinbafrank
qinbafrank|Jun 10, 2026 01:00
Last night, due to the Semianalysis report, which ones were mistakenly killed? The large-scale implementation of CPO has been postponed to 2027-2028 (in fact, it is not considered a delay, according to the rhythm schedule of testing, performance verification, and mass production, it was originally scheduled for 27-28 years. Those who truly understand the industry logic should have expectations for this). The main force that truly delivers performance in the field of optical interconnection in the 26th year is still the 1.6T pluggable optical module (especially the LPO linear direct drive version) and the NPO near package solution. Core logic: The demand for 1.6T and above optical modules is still high. LPO reduces power consumption by removing DSP linear direct drive (30-50% lower than traditional DSP modules), while NPO brings the optical engine closer to ASIC (about 2-7cm distance), achieving performance close to CPO while retaining maintainability. However, the dependence of optical modules (including LPO) and NPOs on upstream core devices is still extremely high: 1) Laser/Light Source: The most certain and direct benefit. LPO/NPO both require high-power, low-noise, and arrayed lasers (VCSEL for short-range MMF, EML/DFB/external laser source ELS for SMF 1.6T+). CPO also uses lasers, but LPO/NPO will increase production first, and the shipment guidance for 2026 will be directly driven. Lumentum, Coherent, Broadcom and other laser manufacturers, as well as the domestic InP/VCSEL supply chain, will benefit first. 2) Photonic Chips (including SiPh and InP based chips): silicon optical modulators MUX/DEMUX、 The detector (PIN/APD) is the core of the 1.6T module/NPO optical engine. LPO linear drive requires higher linearity for optical chips, and the demand is increasing instead of decreasing. Silicon photonics integration can also reduce material usage, but the overall market size is expanding. 3) Optical Engine Assembly: TOSA/ROSA, Optical Engine Integration (Silicon Optical+Laser+Detector). NPO places the light engine on the board, which is closer to modular packaging, with a large quantity and high technical threshold. 4) Testing&Fab Equipment: The demand for high-density testing equipment (high-speed eye diagram testing, probe stations, optoelectronic coupling testing) has exploded. The Bernstein report explicitly states that "testing" will be one of the real sources of income in 2026. MOCVD epitaxial growth equipment and wafer level packaging/alignment equipment also benefit from the ramp up of mass production. 5) Materials: InP substrate (high-speed laser/detector core), GaAs substrate (VCSEL), optical crystals, fiber preform, etc. The supply and demand gap of upstream materials will expand with the increase of 800G/1.6T volume. ABF (Ajinomoto Build up Film) and CCL (Copper Cladd Laminated) high-frequency substrate materials also directly benefit - NPOs require high-performance PCBs/substrates to carry optical engines, and high-frequency circuits inside modules also rely on these. The most wrongly killed are: 1.6T optical module leader, company with strong LPO/NPO capabilities, light source/FAU/testing/PCB/CCL/ABF. Not counting the wrong kills: Pure themed stocks that rely solely on CPO concepts, lack orders, lack customer verification, and do not have 1.6T or LPO/NPO redemption capabilities. Simply put, it means: The CPO delay (which is not actually considered a delay, as some investors had already anticipated it) is not due to the decline of optical interconnection; It is the transition of optical interconnect investment from "long-term imagination" to "2026 performance realization". So the real focus is not on the companies that are the best at telling CPO stories, but on the companies that have already placed orders in 1.6T, LPO/NPO, light sources, testing, and high-speed boards. This is exactly what the Bernstein report emphasizes as the "investment rhythm" - buying the most difficult bottleneck that has already been cashed in cash flow, rather than a pure story. This article is sponsored by @ bitget_zh, titled 'Bitget Buying US Stocks: Instant Entry, Smooth Trading'
+3
Mentioned
Share To

Timeline

HotFlash

APP

X

Telegram

Facebook

Reddit

CopyLink

Hot Reads