My core view was: this semiconductor rally is not over yet, but it is no longer in its early stage.
The first phase was led by core bellwethers like NVDA, AMD, and TSM, which drove index-level confirmation of the trend.
The second phase saw expansion into sub-sectors such as memory, AI optical communications, equipment, testing, and materials.
We are now clearly in the late stage of Phase Two.
If risk appetite in the market remains elevated, the third phase is likely to look like this: capital rotates further out from leading megacaps and mid-cap core names into smaller, higher-beta names with stronger narratives but less fully proven fundamentals.
This piece is about how I personally screen for stocks that are most likely to be driven by the market in this third phase of the semiconductor cycle.
Phase Three is not about randomly buying small caps. The focus should be on names with:
● High beta (I am generally looking at market caps between $1B–$5B)
● Close proximity to the AI thematic core
● Increasing volume and liquidity
● Early-stage relative strength
● Clear catalysts (earnings beats, new orders, data center demand acceleration, product validation, pricing cycles, or institutional inflows)
Below are the key thematic areas most likely to attract attention in the next stage of the AI semiconductor boom:
The logic is straightforward:
As AI data centers scale rapidly, bandwidth bottlenecks between GPU clusters become increasingly severe. This drives demand for optical modules, optical components, silicon photonics, optical engines, and laser systems, all of which are likely to be repriced by the market.
Recent research from Citrini Research also highlights that AI compute demand is pulling optics out of the traditional telecom cycle, positioning optical interconnect as a core beneficiary of AI infrastructure buildouts.
Within this space, the large-cap and mid-cap leaders such as LITE and COHR have already seen significant rerating.
If we enter Phase Three, capital is likely to rotate further into smaller and more speculative optical names.Two small-cap names worth watching are POET and LASR:
● POET: More directly tied to silicon photonics and optical engine platforms. It carries a stronger narrative angle and is well positioned as a “high-conviction AI optical interconnect micro-cap story.”
● LASR: More exposure to laser systems and industrial lasers. The AI optical communication angle is less pure than POET, but it benefits from broader exposure to industrial manufacturing and AI infrastructure expansion, with a more established business base.
This theme is especially suited for Phase Three because it offers a simple, intuitive narrative:
The larger AI data centers become, the more extreme their power demand, and the more important power semiconductors become.
Two names to watch here are WOLF and NVTS:
● WOLF: A high-risk, high-beta SiC turnaround candidate. It has been heavily de-rated due to prior fundamental and cyclical pressures. However, if the market begins to price in AI data center power demand and a semiconductor cycle recovery, its upside convexity could be significant.
● NVTS: More focused on GaN power semiconductors and high-efficiency power architectures. It is closely aligned with AI server power upgrades and is easier for the market to understand thematically.
The advantage of this theme is that the logic is structurally sound and industrially grounded.
The downside is that many of the leading names have already re-rated significantly.Therefore, in Phase Three, the focus is not on already fully priced winners, but on names that still have room for a second acceleration leg.
Key names to monitor include: AEHR, ICHR, UCTT, AXTI, and ATOM.
Unlike traditional semiconductor plays, this theme focuses on real-world enterprise deployment scenarios and monetizable AI solutions, rather than pure chip or optical hardware narratives.
Markets tend to favor a combination of:
● verifiable revenue growth
● high narrative optionality
● technological moat perception
This theme may only fully emerge later in the cycle, but it is worth watching.
Key names include:
● MAAS: B2B AI + distributed compute + AI robotics. Combines compute services, vertical industry models, and localized deployment, along with small lithium charging robotics hardware. It sits at the intersection of enterprise AI deployment and intelligent hardware.
● SOUN: Enterprise voice AI / agent systems. Reported Q1 2026 revenue growth of 52%, driven by automotive, IoT, enterprise, and consumer applications. It fits the “AI entering enterprise interaction layer” narrative.
● INOD: AI data engineering and training data services. Reported Q1 2026 revenue growth of 54%, with full-year guidance raised to 40%+. A clear beneficiary of enterprise AI model infrastructureization.
● SERV: AI delivery robotics. Focused on last-mile delivery robots. Commercialization is still early, but it offers strong narrative-driven upside potential.
The first phase was led by core bellwethers like NVDA, AMD, and TSM, which drove index-level confirmation of the trend.
The second phase saw expansion into sub-sectors such as memory, AI optical communications, equipment, testing, and materials.
We are now clearly in the late stage of Phase Two.
If risk appetite in the market remains elevated, the third phase is likely to look like this: capital rotates further out from leading megacaps and mid-cap core names into smaller, higher-beta names with stronger narratives but less fully proven fundamentals.
This piece is about how I personally screen for stocks that are most likely to be driven by the market in this third phase of the semiconductor cycle.
Phase Three is not about randomly buying small caps. The focus should be on names with:
● High beta (I am generally looking at market caps between $1B–$5B)
● Close proximity to the AI thematic core
● Increasing volume and liquidity
● Early-stage relative strength
● Clear catalysts (earnings beats, new orders, data center demand acceleration, product validation, pricing cycles, or institutional inflows)
Below are the key thematic areas most likely to attract attention in the next stage of the AI semiconductor boom:
Theme 1: Optical Interconnect / Silicon Photonics / Small-Cap Optical Communications
If memory has been one of the strongest fundamental drivers in this semiconductor cycle, then AI optical interconnect is one of the most narrative-driven high-beta themes.The logic is straightforward:
As AI data centers scale rapidly, bandwidth bottlenecks between GPU clusters become increasingly severe. This drives demand for optical modules, optical components, silicon photonics, optical engines, and laser systems, all of which are likely to be repriced by the market.
Recent research from Citrini Research also highlights that AI compute demand is pulling optics out of the traditional telecom cycle, positioning optical interconnect as a core beneficiary of AI infrastructure buildouts.
Within this space, the large-cap and mid-cap leaders such as LITE and COHR have already seen significant rerating.
If we enter Phase Three, capital is likely to rotate further into smaller and more speculative optical names.Two small-cap names worth watching are POET and LASR:
● POET: More directly tied to silicon photonics and optical engine platforms. It carries a stronger narrative angle and is well positioned as a “high-conviction AI optical interconnect micro-cap story.”
● LASR: More exposure to laser systems and industrial lasers. The AI optical communication angle is less pure than POET, but it benefits from broader exposure to industrial manufacturing and AI infrastructure expansion, with a more established business base.
Theme 2: Power Semiconductors / AI Data Center Power Infrastructure
AI data centers are not just about GPUs and HBM. As the cycle matures, the market inevitably shifts attention toward power constraints: electricity supply, cooling systems, power conversion, SiC, GaN, and power management.This theme is especially suited for Phase Three because it offers a simple, intuitive narrative:
The larger AI data centers become, the more extreme their power demand, and the more important power semiconductors become.
Two names to watch here are WOLF and NVTS:
● WOLF: A high-risk, high-beta SiC turnaround candidate. It has been heavily de-rated due to prior fundamental and cyclical pressures. However, if the market begins to price in AI data center power demand and a semiconductor cycle recovery, its upside convexity could be significant.
● NVTS: More focused on GaN power semiconductors and high-efficiency power architectures. It is closely aligned with AI server power upgrades and is easier for the market to understand thematically.
Theme 3: Secondary Expansion in Equipment, Testing, and Materials
If the semiconductor cycle continues to strengthen, equipment, testing, and materials will inevitably see renewed capital rotation.The advantage of this theme is that the logic is structurally sound and industrially grounded.
The downside is that many of the leading names have already re-rated significantly.Therefore, in Phase Three, the focus is not on already fully priced winners, but on names that still have room for a second acceleration leg.
Key names to monitor include: AEHR, ICHR, UCTT, AXTI, and ATOM.
Ticker | Rationale |
| AXTI | Positioned across materials, optical communications, and compound semiconductors; narrative-friendly and easy for the market to reframe |
| AEHR | High-beta test equipment name; tends to be aggressively bid when semiconductor equipment sentiment expands |
| ATOM | Micro-cap with high volatility; fits classic Phase Three “sentiment-driven small-cap” profile |
| ICHR | Beneficiary of equipment cycle recovery and order improvement, more cyclical and fundamentals-driven |
Theme 4: B2B AI Commercialization + AI Robotics
In the third phase of the semiconductor rally, if capital continues searching for peripheral high-beta opportunities, an increasingly important theme is enterprise AI adoption and intelligent hardware applications.Unlike traditional semiconductor plays, this theme focuses on real-world enterprise deployment scenarios and monetizable AI solutions, rather than pure chip or optical hardware narratives.
Markets tend to favor a combination of:
● verifiable revenue growth
● high narrative optionality
● technological moat perception
This theme may only fully emerge later in the cycle, but it is worth watching.
Key names include:
● MAAS: B2B AI + distributed compute + AI robotics. Combines compute services, vertical industry models, and localized deployment, along with small lithium charging robotics hardware. It sits at the intersection of enterprise AI deployment and intelligent hardware.
● SOUN: Enterprise voice AI / agent systems. Reported Q1 2026 revenue growth of 52%, driven by automotive, IoT, enterprise, and consumer applications. It fits the “AI entering enterprise interaction layer” narrative.
● INOD: AI data engineering and training data services. Reported Q1 2026 revenue growth of 54%, with full-year guidance raised to 40%+. A clear beneficiary of enterprise AI model infrastructureization.
● SERV: AI delivery robotics. Focused on last-mile delivery robots. Commercialization is still early, but it offers strong narrative-driven upside potential.
