What Are the CPO Concept Stocks? The U.S. Stock Leaders’ Landscape and the Logic Behind AI Computing Infrastructure

In early June 2026, the CPO (Co-packaged Optical) sector experienced synchronized high-level fluctuations in both the A-share and U.S. stock markets. On June 1st, the A-share CPO concept sector declined by 3.79% in a single day, with net main capital outflows of 9.83B yuan; however, over a longer cycle, this sector has gained over 200% since April 2025, with a total market capitalization surpassing 3.3 trillion yuan. Behind the volatility, the market's debate over CPO's technical route divergence, valuation digestion pressure, and the long-term certainty of AI computing power is intensifying.

The real question worth asking is not "How much longer can CPO keep rising," but rather: What kind of structural changes are occurring in the value distribution of the CPO industry? How do the positions of core U.S. stock leaders and A-share targets differ within the industry chain? When conducting cross-market allocations via platforms like Gate, what core logic should be the focus?

Why CPO is Moving from Concept to Performance Realization

CPO technology directly packages optical engines and switching chips, shortening optical paths and reducing power consumption, and is regarded as a key solution to the interconnection bottleneck of AI clusters with millions of cards. The industry generally defines 2026 as the "Year of CPO Industry Commercialization," but a more accurate statement is: upstream optical chip segments entered a performance explosion phase as early as late 2025, while large-scale deployment of system-level CPO still awaits the adoption of NVIDIA's Rubin Ultra platform in 2027.

The driving force behind the CPO sector has shifted from "theme speculation" to "performance verification." In Q1 2026, Lumentum's revenue surged 90% year-over-year, with a gross margin of 47.9%; Coherent's data center business revenue reached $1.36 billion, up 39% YoY; Fabrinet's revenue hit a record $33k. The order visibility for these three companies generally extends to 2028, with indium phosphide substrates experiencing a supply-demand gap exceeding 70%, and prices soaring from $800 per piece to $2,500.

In the short term, the industry faces dual pressures of "capacity bottlenecks" and "technological route competition." The LPO (Laser-Photonic-Optical) solution poses a short-term threat to CPO in terms of power consumption and cost, with Microsoft having already adopted LPO first. However, in the long run, the CPO market size in vertical expansion scenarios is expected to reach $103.9 billion by 2030, more than four times that of lateral expansion scenarios. NVIDIA's adoption of Scale-Up CPO on the Rubin Ultra platform in 2027 will be the most certain growth inflection point for the industry.

CPO is reshaping the value distribution of the optical communication industry chain. In the traditional optical module era, profits were concentrated in module assembly; in the CPO era, value shifts upstream to optical chips (such as EML, CW lasers) and advanced packaging. Lumentum leverages its monopoly in EML chips to dominate upstream profits; Coherent achieves vertical integration through its three major photonic platforms (InP, silicon photonics, VCSEL); Fabrinet, as a pure OEM, acts as a "seller of shovels."

Breakdown of the Four Core U.S. CPO Leaders

Currently, in the U.S. CPO industry chain, the positioning and competitive barriers of four companies differ significantly.

Lumentum: Absolute monopolist in EML chips. Over 90% of the 200G EML chips required for 1.6T optical modules come from Lumentum. NVIDIA invested $2 billion in Lumentum in March 2026 to lock in future capacity. The company's OCS (Optical Circuit Switch) backlog has exceeded $400 million, with multi-year multi-billion dollar agreements signed. In FY2026 Q3, revenue was $808 million, up 90% YoY, with a gross margin of 47.9%, and high-end EML chip orders extend to 2028.

Coherent: Vertical integration across three major photonic platforms. The only company worldwide that controls indium phosphide, silicon photonics, and VCSELs simultaneously. NVIDIA also invested $2 billion here. In FY2026 Q1, data center revenue reached $1.36 billion, accounting for 75.4% of total revenue. The company has disclosed a $15 billion market for CPO services, with a 6.4T CPO solution showcased at OFC 2026.

Marvell: Dual-track layout of AI chips and optical interconnects. NVIDIA invested $2 billion here as well. Marvell acquired Celestial AI for $5.5 billion to obtain CPO optical interconnect technology and is now developing its own CPO ASIC. The company significantly raised revenue forecasts for FY2027 and FY2028, with broker estimates of about 40% YoY growth in FY2027, and interconnect product revenue increasing over 70%.

Fabrinet: A highly certain OEM logic. Does not develop optical technology independently but manufactures complex optical modules for companies like COHR and LITE. In Q3 FY2026, revenue hit a record $1.21B, up 39%. Management confirmed early CPO revenues, working with three clients to advance CPO projects.

The competitive landscape among these four companies features "upstream chip oligopoly, midstream foundry benefiting from any winner, and system integrators relying on AI chips bundling." Lumentum's EML barrier is the highest, Fabrinet has the smallest risk exposure, Marvell is the most flexible but also faces the highest technological route uncertainty.

How the Technical Route Dispute Between CPO and LPO Affects the Market

The main market debate on CPO currently centers on whether LPO (Laser-Photonic-Optical) will capture a larger share during the transition period, thereby delaying CPO's ramp-up.

LPO eliminates DSP chips in traditional optical modules, reducing power consumption by about 50%, and Microsoft has already adopted it as the main technical route. AAOI is a core beneficiary of the LPO route, with Q1 2026 revenue hitting a record $151.1 million, up 51% YoY. The company has provided a revenue guidance of $1 billion for 2026 and increased its monthly capacity target for 800G/1.6T to 650k units.

The two technologies are not entirely substitutive. LPO has cost advantages in short-distance, low-port-density scenarios; CPO is irreplaceable in ultra-large-scale clusters and vertical expansion scenarios. Bernstein's 97-page deep report indicates that large-scale adoption of CPO will be difficult before 2028, with more certain performance increments in LPO/NPO segments in 2026.

The coexistence of both routes is expected to continue until around 2028. Investors should distinguish: Lumentum and Coherent benefit from both CPO and traditional optical module upgrades (with demand for 800G/1.6T still exploding); while AAOI's valuation implies an overly optimistic LPO scenario, with current Non-GAAP gross margins of only 29.2%, indicating weaker profitability compared to peers.

How Institutional Funds and Liquidity Are Repricing the CPO Sector

From late 2025 to May 2026, there was a clear rotation of funds in the U.S. optical communications sector: some capital shifted from well-priced AI hardware giants like NVIDIA to optical network companies.

Institutional actions: In Q1 2026, in the A-share CPO sector, Zhongtian Technology was increased by institutions by 182 million shares; Hengtong Optoelectronics saw institutional holdings rise from 40.72% to 43.94% of circulating shares. In the U.S., after NVIDIA's investment announcement, institutional holdings in Coherent and Lumentum increased significantly. J.P. Morgan issued a "Neutral" rating on Fabrinet, believing current stock prices partly reflect the super-cycle benefits of AI hardware.

The pricing power of the CPO sector is shifting from retail sentiment to institutional performance verification. The 2025 rally relied more on concept expectations; after Q1 2026 earnings, the market began to re-evaluate targets based on "order visibility, gross margin trends, capacity expansion progress," and other hard metrics. Stocks lacking technological barriers and performance support declined more sharply during the June 1st correction.

If the Federal Reserve enters a rate-cut cycle in the second half of the year, easing global liquidity will further benefit tech growth stocks, and the CPO sector may see increased capital allocation. However, sector differentiation will intensify: upstream optical chips (Lumentum, Coherent) and foundries (Fabrinet) with more certain earnings will face less valuation digestion pressure; companies relying on single clients or a single technical route will face higher volatility risks.

Pathways and Precautions for US Stock CPO Allocation via Gate

On June 1, 2026, Gate officially launched stock trading services, allowing users to directly trade U.S. stock CPO core targets using USDT/USDC.

Gate stocks are directly connected to major markets like NYSE and Nasdaq. Currently, it supports intraday market price trading of CPO-related targets such as COHR, LITE, MRVL, FN, AAOI, without involving funding rates or overnight holding fees. Additionally, Gate's perpetual contract section for stocks has launched AAOI and other products.

The launch of Gate stock trading reduces friction costs for crypto users seeking U.S. stock assets. Investors interested in the CPO track can now perform cross-market allocations within the same account without leaving the crypto ecosystem. However, it is important to note that U.S. stock trading hours, liquidity depth, company fundamentals tracking, and other factors differ significantly from traditional crypto assets.

The CPO industry still faces multiple risks, including technological route uncertainty, capacity bottlenecks, and US-China tech friction. On June 1, 2026, the sector declined 3.79% in a single day, with net main capital outflows of 1.6T yuan, reflecting high-level game characteristics. Investors should distinguish between industry leaders with technological barriers and performance support, and concept-driven hype stocks.

FAQ

What are CPO concept stocks?

CPO concept stocks are listed companies involved in the co-packaged optical technology industry chain, including upstream optical chips (like Lumentum), midstream optical modules (like Coherent), and downstream OEM packaging (like Fabrinet).

Who are the leading U.S. CPO companies?

The main U.S. CPO leaders include Lumentum (monopoly in EML chips), Coherent (vertical integration of three major photonic platforms), Marvell (AI chips + optical interconnects), Fabrinet (pure OEM), and AAOI (representative of the LPO route).

Which technical route has better prospects, CPO or LPO?

CPO is irreplaceable in ultra-large-scale clusters and vertical expansion scenarios, but LPO has short-term cost advantages in short-distance applications. Both are expected to coexist until 2028, with NVIDIA's 2027 Rubin platform adoption of CPO being a key inflection point.

How is the performance realization of CPO concept stocks?

In Q1 2026, Lumentum's revenue surged 90% YoY, Coherent's data center revenue grew 39%, and Fabrinet's revenue increased 39%, hitting record highs. Upstream optical chip segments have already entered a performance explosion phase.

What are the core risks in the CPO industry?

Key risks include technological route uncertainty (LPO competition), capacity bottlenecks (indium phosphide substrate shortages), valuation pressures at high levels, and potential impacts from US-China tech frictions on the supply chain.

How does Gate support trading of U.S. stock CPO concept stocks?

Gate's stock trading service supports targets like COHR, LITE, MRVL, FN, AAOI, allowing users to buy stocks directly with USDT/USDC, without funding rates or overnight fees.

Where is the future growth inflection point for the CPO sector?

Industry consensus points to 2027, when NVIDIA's Rubin Ultra platform first adopts Scale-Up CPO, as the most certain growth inflection point. The vertical expansion CPO market is expected to exceed $100 billion by 2030.

How to evaluate the investment value of CPO concept stocks?

Focus on companies' technological barriers (self-developed optical chips), customer loyalty (entry into NVIDIA/cloud vendor supply chains), capacity expansion progress, and gross margin trends, rather than short-term stock price increases.