Taiwanese Nodes on the AI Industry Chain: 9 Taiwanese Stock Picks Favored by 'New Stock God' Serenity

marsbitPublished on 2026-06-09Last updated on 2026-06-09

Abstract

Investment analyst Serenity has spotlighted 9 Taiwanese stocks positioned within three key themes of the AI hardware supply chain: CPO (Co-packaged Optics), ASIC (Application-Specific Integrated Circuits), and Compound Semiconductors. For the CPO theme, anchored by TSMC's COUPE platform expected to mass produce in 2026, Serenity highlighted five stocks: FOCI (fiber array unit supplier), Shunsin Technology (optical packaging & testing), Xintec (wafer-level packaging & testing), MSSCorps (inspection/measurement tools), and Nextronics (connectors & thermal modules). Regarding the ASIC theme, driven by hyperscalers' custom chips, three companies were named: Alchip (design services for AWS), Unimicron (ABF substrates for Google TPU/AWS), and MediaTek (potential collaboration on Google TPU). A singular, strong conviction pick is Win Semiconductor, a compound semiconductor foundry and leader in InP/GaAs, seen as a beneficiary in areas like laser sources for silicon photonics and LEO satellite communications. Serenity argues that current valuations are not in a bubble, geopolitical risks are overstated, and the real systemic risk lies not in cross-strait tensions but in potential future cuts to hyperscaler (Microsoft, Amazon, Google, Meta, Oracle) capital expenditure—the primary driver of the current AI infrastructure boom.

Author: Ada, Deep Tide TechFlow

The "New Stock God" Serenity, popular on X, recently shared his systematic views on Taiwan's AI industry chain. He named 9 Taiwanese stocks covering three main themes: CPO (Co-Packaged Optics), ASIC (Application-Specific Integrated Circuit), and compound semiconductors. He identified CPO as "the biggest industrial theme for the next phase." Goldman Sachs predicts the global Optical TAM will explode from $15 billion in 2026 to $154 billion in 2028, with CPO's share growing from $164 million to $91 billion. He also explicitly stated: Taiwan's AI industry chain does not have a bubble; the real risk is not cross-strait relations, but the capital expenditures of hyperscale cloud service providers.

TSMC COUPE Countdown to 2026 Mass Production, 5 Taiwanese Stocks Cover Each Segment

The hard timeline anchoring Serenity's CPO thesis is TSMC's COUPE (Compact Universal Photonic Engine) platform entering mass production in 2026.

According to TrendForce's April 1st report citing TSMC's Advanced Packaging Integration Director Hou Shangyu at SEMICON Taiwan, COUPE, based on SoIC technology for heterogeneous integration of electronic and photonic integrated circuits, is on schedule for mass production in 2026. Hou also identified three bottlenecks for CPO scaling: wafer-level testing, Fiber Array Unit (FAU) integration, and high-speed optical packaging assembly. The 5 CPO Taiwanese stocks named by Serenity correspond precisely to these three bottleneck segments and their surrounding support infrastructure.

The first is FOCI (First-O-Link Optronics, 3363.TWO), with a market cap of approximately $2.8 billion. FOCI is a key partner for NVIDIA and TSMC in the FAU field and a core supplier within the COUPE architecture. Serenity stated on X that FOCI's current market cap of around $2.8 billion "has not fully reflected its strategic position in the industry chain." Morgan Stanley estimates FOCI's FAU business could contribute approximately NT$20 billion in revenue from a single product line by 2028, whereas it is currently nearly zero. As early as March, Hunterbrook Media and Citrini Research, through cross-referencing four patents from FOCI, Himax, and TSMC, confirmed a "fingerprint-level" correspondence between the three companies on 22-channel FAUs, providing the strongest external evidence of FOCI's entry into COUPE's core supply chain.

The second is Shunsin Technology (6451.TWSE), with a market cap of approximately $1.4 billion. Shunsin is the optical communication packaging and testing subsidiary of Foxconn Group. Serenity once listed Shunsin as "Taiwan NVDA CPO supply chain idea #1" on X, reasoning that "Foxconn is NVIDIA's ODM, akin to Celestial getting a free ride with MRVL's listing." He provided a personal estimate that Shunsin's forward P/E for 2027 is around 20x.

The third is Xintec (3374.TWO), with a market cap of approximately NT$47.6 billion. Xintec is TSMC's wafer-level packaging and testing subsidiary, with Chairman CH Chen. Its main businesses are wafer-level chip-scale packaging and wafer testing. According to a DIGITIMES report on May 29, 2026, Xintec is preparing capacity for testing business expansion in the second half of 2026. In a recent interview, Serenity categorized Xintec as a "beneficiary of COUPE-related testing business."

The fourth is MSSCorps (6830.TWO), with a market cap of approximately $1.4 billion. Serenity explicitly stated on X that he had established a position in MSSCorps, offering a strong judgment: "This looks like a functional monopoly in CPO inspection. The market might confuse it with material/failure analysis oligopolies like MA-tek and iST." He then cited MSSCorps' own statement, "The company aims to capture 90% of the CPO inspection market share," to support this. MSSCorps' customer mapping includes TSMC, NVIDIA, AAPL, AMAT, etc. Among the three CPO bottlenecks pointed out by TSMC, "wafer-level testing" is the segment where MSSCorps directly holds a key position.

The fifth is Nextronics (8147.TWO), with a market cap of approximately $210 million. Serenity's reason for establishing a position in Nextronics is its supply of "CPO connectors and Cage Thermal Modules" to NVIDIA's CPO supply chain. Nextronics has the smallest market cap among the 5 CPO Taiwanese stocks, fitting Serenity's typical stock selection profile of "small market cap + no coverage + positioned at a physical bottleneck."

Roughly sorted by supply chain segment, the 5 stocks are: FAU access (FOCI), optical communication packaging & testing (Shunsin Technology, Xintec), yield inspection (MSSCorps), connectors & thermal management (Nextronics). The downstream customers for this entire chain are highly concentrated in NVIDIA and TSMC.

Three Taiwanese ASIC Companies Bind to Three Hyperscalers' Custom Chip Development

Beyond NVIDIA GPUs, the continued volume ramp of hyperscale cloud service providers' self-developed ASICs, such as AWS Trainium, Google TPU, and Microsoft Maia, constitutes the second main theme in Serenity's Taiwanese stock list.

Alchip (3661.TWSE) is the most representative stock on this line. Founded in 2003 and headquartered in Taipei, Alchip specializes in advanced CMOS ASIC design and manufacturing. According to industry chain data compiled by Global Tech Research, Alchip primarily assists Amazon's Annapurna Labs with backend design for the AWS Trainium and Inferentia chip series. Current projects include Trainium 1 and Inferentia 2. Serenity recently stated that Alchip is expected to gain more market share in the design of AWS Trainium 3, especially as Amazon's recent private investment in Alchip is interpreted by the market as a significant signal. Alchip's collaboration with Ayar Labs in the CPO field also represents further expansion of its potential market.

Unimicron (3037.TWSE) is an ASIC beneficiary from the perspective of ABF substrates and PCBs. According to Gotrade News citing Bernstein's May 4th research report, Unimicron is expected to secure about 35% of the ABF substrate share for NVIDIA's high-end GPUs and over 50% share for ASIC chips like Google TPU and AWS Trainium. Bernstein listed TSMC and Unimicron as "the two most representative Taiwanese stocks in the Asian market's AI Capex cycle." Unimicron's Q1 2026 revenue grew 8% quarter-over-quarter, with a gross margin of 18%.

News of MediaTek's (2454.TWSE) collaboration with Google on TPUs is already widely circulated in the market. According to Global Tech Research analysis, MediaTek may participate in providing backend design support for future Google TPU V7e/V7p. Serenity recently noted that while some of this positive news is already partially reflected in the stock price, he believes MediaTek's future ASIC business growth is still worth continuous attention. In the coming years, MediaTek has the opportunity to become an extremely important part of the hyperscale supply chain for major US cloud providers.

These three Taiwanese companies bind to the core segments of hyperscalers' self-developed chips from three different dimensions: design services, substrate packaging, and SoC design.

Win: The Undervalued InP/GaAs Duopoly

Serenity's most resolute and explicitly stated holding among Taiwanese stocks is Win Semiconductor (3105.TWO).

Win is one of the world's leading compound semiconductor foundries, forming a duopoly with GlobalFoundries in InP (Indium Phosphide) and GaAs (Gallium Arsenide) foundry services. The market's traditional impression of Win comes from its relationship with the SpaceX Starlink and Broadcom (AVGO) supply chains.

Serenity's recent statement on X was: "Win will win. So I am long Win." Serenity's core judgment is that Win is one half of the InP and GaAs foundry duopoly. Future products for LEO (Low Earth Orbit) satellites like SpaceX Starlink, Continuous Wave (CW) lasers, and vision/radar light sources for humanoid robots will all utilize Win's foundry capacity.

He provided a valuation estimate: Win's forward P/E for 2027 is around 35x. While this may not seem cheap, he believes "management is intentionally guiding earnings low; the P/E will appear extremely cheap after actual profit realization." Win's business of assisting silicon photonics companies in expanding laser mass production capabilities is the future core growth engine in Serenity's view, and this part is not yet fully priced in by the market.

InP is the segment where Serenity first established a position in a US stock, AXTI, which he termed a "Hormuz Strait" type of investment. He compares InP substrates to "the oil of the future AI optical communication era." Win's position in Taiwanese stocks is the downstream foundry node of this industry chain.

Segments Serenity Did Not Mention But Essential for the Full Picture: Upstream Materials, Energy, Liquid Cooling

Serenity's 9 named Taiwanese stocks are concentrated in the midstream segments of CPO and ASIC. For a complete view of Taiwan's AI industry chain, upstream materials and downstream energy/liquid cooling sectors are equally indispensable. Serenity did not directly cover this part in his public remarks, but it cannot be omitted for a full understanding.

On the upstream materials side, TSMC itself (TSMC, 2330.TWSE) is the central node of the entire picture. According to Bernstein forecasts, TSMC's earnings will grow about 40% in 2026, with a compound annual growth rate of about 20% for 2027-2028; the company has set its 2026 capital expenditure guidance at the high end of the $52-56 billion range. In the upstream transmission chain of AI Capex, TSMC is both an amplifier and a throttle. Its capital expenditure direction directly determines the order rhythm for all downstream Taiwanese stocks in packaging & testing, substrates, inspection, connectors, etc.

On the energy and liquid cooling side, Taiwanese AI data centers are facing tangible power supply pressure. According to Reccessary's December 2025 report citing Taipower System Planning Director Yi Xuquan, global key grid equipment like transformers and gas turbines are in short supply. Gas turbine delivery lead times have extended from 2-3 years to 7-8 years, with prices doubling, and "grid construction speed simply cannot keep up." This supply-side bottleneck has created structural opportunities for three key Taiwanese stocks.

First, Delta Electronics (2308.TWSE), as the power management and integrated liquid cooling solution leader, has penetrated both the white space (IT computing) and grey space (power & cooling) scenarios of AI data centers. Second, AVC (3017.TWSE) is a global leader in thermal solutions, covering a full stack of technologies from cold plate liquid cooling to immersion cooling. Lastly, Auras Technology (3324.TWSE) focuses on liquid cooling module manufacturing and is a tier-1 supplier to NVIDIA. According to Goldman Sachs predictions, the global server cooling total addressable market will grow 111%, 77%, and 26% in 2025-2027, respectively, reaching $17.6 billion in 2027. This growth curve is synchronized with the CPO growth curve within the AI capital expenditure cycle.

The Real Risk is Not Cross-Strait Relations, But Hyperscale Cloud Service Providers' Capital Expenditures

Serenity's judgment on the overall risk of Taiwan's AI industry chain differs from the mainstream market narrative.

He recently stated that Taiwan's AI supply chain currently does not have a bubble, and the valuations of many Taiwanese companies are still relatively low compared to their US Nasdaq counterparts; geopolitical risks are overly amplified by the market because Taiwan is already an indispensable part of the global technology supply chain, and "no country would choose a zero-sum, large-scale supply chain disruption." Looking 7-10 years ahead, the US is building a domestic supply chain through companies like Intel, but there is no particularly significant systemic risk in the next 2-3 years.

He pointed out that the real risk worth tracking is the capital expenditures of hyperscale cloud service providers. The continuous expansion of investments by large cloud providers like Microsoft, Amazon, Google, Meta, and Oracle is the core driver of the current AI industry boom. "As long as these companies continue to increase AI infrastructure construction, Taiwan's supply chain will continue to benefit; but if hyperscale cloud service providers start cutting capital expenditures, Taiwan's AI supply chain may experience a relatively significant valuation adjustment."

Initial signs of a "stress test" for this judgment are beginning to appear.

Recent marginal changes in the market are worth noting as tracking clues. On June 3rd, Broadcom announced its Q2 FY2026 results. The Q3 FY2026 guidance for AI chip revenue was $16 billion, approximately $1.2 billion below the LSEG-compiled sell-side consensus estimate of $17.2 billion. Broadcom CEO Hock Tan also maintained the full-year AI semiconductor revenue guidance of $56 billion without raising it. The market interpreted the "lack of an upward revision" negatively. On June 5th, the Philadelphia Semiconductor Index (SOX) plunged 10.26%, its largest single-day drop since March 2020. On the same day, South Korea's KOSPI fell 5.54%, and on June 8th, KOSPI opened down 8.37%, triggering a trading halt.

However, the key reference point to watch currently is TSMC's 2026 capital expenditure, locked at the high end of the $52-56 billion range, showing no substantial reduction yet. The capital expenditure guidance from hyperscale cloud service providers themselves also remains on an upward revision track. Whether Serenity's predicted "real risk" is triggered depends on the capital expenditure guidance from Microsoft, Amazon, Google, Meta, and Oracle in their Q2 and Q3 2026 earnings reports. This is the core premise for the validity of Serenity's Taiwanese stock list and the most important marginal variable to track in the next quarter.

Related Questions

QAccording to the article, what are the three main sub-sectors of the Taiwanese AI industrial chain that Serenity's highlighted stocks cover?

AThe three main sub-sectors are CPO (Co-packaged Optics), ASIC (Application-Specific Integrated Circuit), and compound semiconductors.

QWhat is the key time anchor supporting Serenity's focus on the CPO sector, and which TSMC platform is central to this timeline?

AThe key time anchor is TSMC's COUPE (Compact Universal Photonic Engine) platform, which is scheduled to enter mass production in 2026.

QIn Serenity's view, where does the real risk to the Taiwanese AI supply chain lie, not in geopolitics?

ASerenity believes the real risk lies in the capital expenditure trends of hyperscale cloud service providers (like Microsoft, Amazon, Google, Meta, Oracle). A cutback in their AI infrastructure spending could lead to significant valuation adjustments for the Taiwanese supply chain.

QName the three Taiwanese companies mentioned in the ASIC segment and the major hyperscalers they are respectively linked to.

A1. Alchip (Worldsemi-KY) is linked to AWS (Amazon Web Services) for its Trainium/Inferentia chips. 2. Unimicron (欣兴电子) is linked to Google TPU and AWS Trainium for ABF substrates. 3. MediaTek is linked to Google for potential involvement in TPU backend design support.

QWhich company does Serenity hold as his most firm and explicitly stated position in Taiwanese stocks, and what is its core business?

ASerenity's most firm position is in Win Semiconductor (稳懋). It is a leading compound semiconductor foundry, a duopoly in InP (Indium Phosphide) and GaAs (Gallium Arsenide) wafer manufacturing.

Related Reads

Sharplink CEO: One Million Ethereum Developers, Who Can Compete?

Etherean Ecosystem: One Million Developers and a Formidable Moat The Ethereum network has surpassed a significant milestone: over one million unique lifetime developers, with approximately 232,000 active in the past year. This vast and growing talent pool is Ethereum's core advantage, far more critical than transient metrics like speed or transaction fees. The central question is not which blockchain is fastest, but where the best builders choose to build long-term. Ethereum's answer lies in a decade-long accumulation of infrastructure, standards, tools, liquidity, and collaborative culture that is exceptionally difficult to replicate. It has become the default operating system for programmable finance. This massive developer base is actively working on complex, high-risk challenges that deepen Ethereum's strategic moat: * **Glamsterdam Upgrade (2026):** Focused on core protocol scalability (ePBS, parallel execution) while preserving core values like credible neutrality. * **Synchronous Composability:** Aims to make numerous Rollups interoperate like a single chain, directly addressing fragmentation concerns. * **Quantum Resistance:** Ethereum leads mainstream ecosystems in coordinated preparation for post-quantum cryptography, with a targeted migration plan around 2029. This developer advantage is self-reinforcing, fueled by: * **Deep Composability:** Applications interact like interoperable financial Lego bricks via shared standards (e.g., EVM, Solidity). * **Credible Neutrality:** Secured by over 900,000 validators, making it trusted by major institutions. * **Modularity:** Rollups (Base, Arbitrium, etc.) expand, rather than fracture, the ecosystem into a tightly connected modular internet economy. * **Culture:** Attracts top-tier researchers and standard-setters who guide the entire industry. In essence, while other chains generate activity, Ethereum is consolidating as the trusted, long-term coordination layer for internet-native finance. Its future is being built now by the architects of the next-generation financial infrastructure.

Odaily星球日报28m ago

Sharplink CEO: One Million Ethereum Developers, Who Can Compete?

Odaily星球日报28m ago

Ethereum Reaches the Milestone of One Million Developers, Sharplink CEO Delves Deep into Ethereum's Future Possibilities

Ethereum Surpasses One Million Developers Milestone: A Look at Its Unshakeable Dominance and Future Joseph Chalom, CEO of Sharplink, reflects on his recent Asia trip where he engaged deeply with Ethereum developers and ecosystem leaders. The most striking takeaway was not just the industry's vibrancy, but the rigorous, long-term vision of local builders. This context brings to life a pivotal statistic: Ethereum has now surpassed one million cumulative developers (1,012,824), with approximately 232,000 remaining active in the past year—a talent pool unmatched by any other crypto ecosystem. The critical question isn't which blockchain is fastest, but where top developers choose to build long-term. Ethereum's answer is unequivocal. Its decade-long lead stems from a unique convergence of technology, institutional culture, economic systems, and composability, cementing its role as the foundational operating system for programmable finance. This massive developer base is tackling the industry's hardest problems, continuously strengthening Ethereum's moat. Key initiatives include: * **The Glamsterdam Upgrade (planned 2026):** Introducing ePBS and Block-level Access Lists for parallel execution and higher throughput while preserving core values like credible neutrality and fair MEV distribution. * **Synchronous Composability:** Projects are working to enable atomic transactions across dozens of Rollups, making them function as one unified chain and eliminating ecosystem fragmentation. * **Post-Quantum Security:** Ethereum is far ahead in preparing for quantum computing threats, with a dedicated foundation working group and testnets targeting a full migration by ~2029—a crucial factor for institutional adoption. Beyond developers, Ethereum's core network effects are its unparalleled composability and unified standards (like EVM and Solidity), which create a powerful flywheel: more developers → better tools → greater liquidity → more institutional participation. Its other decisive advantages include credible neutrality (over 900k validators), a secure modular architecture with interconnected Rollups, and a deeply entrenched culture shaped by top-tier researchers. Ultimately, there's a vast difference between generating short-term activity and becoming the trusted, long-term coordination layer for global native finance. Major institutions prioritize security, trust, and liquidity—areas where Ethereum holds dominant mindshare. The industry's trajectory shows resources consolidating around unified standards, deep liquidity, and developer consensus. After meeting the builders in Seoul and Hong Kong, Chalom is more convinced than ever: Ethereum's unshakeable future is being built right now.

Foresight News39m ago

Ethereum Reaches the Milestone of One Million Developers, Sharplink CEO Delves Deep into Ethereum's Future Possibilities

Foresight News39m ago

Saylor's Latest Long Read: Bitcoin is Not Money, It's Digital Capital, and Money is Built Upon It

Michael Saylor presents his "Digital Asset Stack" theory, positioning Bitcoin as the foundational layer of digital capital. He argues Bitcoin itself should remain unchanged—no staking, inflation, or protocol alterations. Instead, a five-layer financial architecture should be built atop it: Digital Capital (BTC), Digital Credit (e.g., yield instruments like STRC), Digital Currency (stable, yield-bearing instruments pegged to fiat), Digital Yield (leveraged/structured products), and Digital Equity (e.g., MSTR stock, absorbing residual volatility). Saylor asserts this stack transforms Bitcoin's high-volatility, high-energy capital into tailored products: stable currencies for payments/savings, yield instruments for income seekers, and equity for growth investors. This approach meets diverse needs—corporate treasuries, banks, retirees, emerging market users—without compromising Bitcoin's core properties (scarcity, decentralization). The "killer use case" is rebuilding global money, credit, and capital markets on Bitcoin, bridging the fiat world with a superior digital asset foundation. The system leverages traditional finance principles (risk layering, structured products) while using Bitcoin as the ultimate collateral. This expands Bitcoin's utility, drives adoption, and offers a better monetary experience: digital, yield-bearing, stable-value tools for everyday use.

marsbit51m ago

Saylor's Latest Long Read: Bitcoin is Not Money, It's Digital Capital, and Money is Built Upon It

marsbit51m ago

Can SK Hynix's Stock Double Again in This Rally?

The article discusses the highly optimistic price target of approximately $3,500 for SK Hynix stock, set by Aletheia Capital. This target is significantly above the consensus range of $2,000-$2,520 from major brokerages. The core debate is whether SK Hynix deserves a fundamental re-rating beyond its traditional cyclical discount, based on the long-term impact of AI-driven demand. The $3,500 target hinges on three key assumptions holding simultaneously until at least 2027: 1) Continued shortage and high pricing for HBM (High Bandwidth Memory), a critical component for AI chips; 2) Sustained high prices for standard DRAM, as HBM production consumes capacity and constrains general supply; and 3) Strong AI server demand generating substantial, above-expectation free cash flow. SK Hynix's leading ~58% market share in HBM and its early certification with key clients like Nvidia provide a competitive advantage, allowing it to capture significant supply chain premiums. The HBM shortage is seen not just as a niche growth driver but as a catalyst that amplifies profitability across the entire memory business by tightening overall DRAM supply. However, the article cautions that this target represents an optimistic "tail scenario." Key risks include potential supply increases from competitors (Samsung, Micron) by 2027, a possible slowdown in HBM price growth, and high capital expenditures that could erode the projected free cash flow. The divergence in analyst targets reflects the market's uncertainty over whether the current AI-driven boom will temporarily elevate earnings or permanently raise the memory industry's profit baseline.

marsbit55m ago

Can SK Hynix's Stock Double Again in This Rally?

marsbit55m ago

Trading

Spot
Futures

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片