Recently, Siok Technology completed the industrial and commercial changes for a new round of capital increase, with the total scale of this round of financing reaching CNY 220 million. Among them, Tztek Technology (688003.SH) invested CNY 135 million to subscribe for the newly added registered capital, increasing its shareholding ratio from 12.10% to 17.03%. According to the announcement, the pre-money valuation of this round was CNY 1.98 billion, and after adding the CNY 220 million capital increase, the post-money valuation is approximately CNY 2.2 billion.
The investment is targeted at Siok Technology's bright-field nano-patterned wafer defect inspection equipment, the TB2000, which is currently undergoing validation at leading wafer fabs and targets the 14nm to 28nm nodes. Tztek Technology itself is a listed company engaged in industrial vision and precision measurement; this move aligns well with Siok Technology's technological foundation migrated from industrial vision.
The financial figures are actually just a few lines: the full-year revenue for 2025 was CNY 28,900, with a net loss of CNY 82.4252 million for the year; by the end of April this year, total assets were CNY 412 million, net assets were CNY 204 million, and orders on hand amounted to CNY 68.87 million. The revenue is still very small and the company remains in the investment phase, but the valuation is not low, with the reason entirely resting on that piece of equipment.
| Bright-Field Inspection: Universally Recognized as the Toughest Challenge
Wafer defect inspection is roughly divided into three tiers based on the target and technical route. The most basic tier is unpatterned (bare) wafer defect inspection, which examines blank wafers before lithography; it has low difficulty, and domestic products are already shipped in batches. The next tier is dark-field nano-patterned defect inspection, which examines patterned wafers and captures defects via scattered light; it has moderate difficulty, and multiple domestic companies have entered validation or small-batch production. The most difficult is bright-field nano-patterned wafer defect inspection, which also examines patterned wafers but captures defects through imaging comparison; most domestic products are still in the validation stage without scaled mass production.
What makes bright-field inspection so difficult? In one sentence: on a wafer already covered with dense, repetitive patterns, it is necessary to separate true defects from the normal structures of the patterns themselves.
Dark-field inspection relies on the optical signals scattered by defects, making it relatively "immune" to the pattern background and easier to implement. Bright-field inspection involves direct imaging, where defect signals and pattern noise are mixed together, equivalent to finding the one anomaly within repetitive patterns. The difficulties are concentrated in three areas:
First, the optical system. It requires objective lens groups with broad spectral bands, large fields of view, and high resolution, often necessitating the deep ultraviolet (DUV) band. Aberrations, illumination uniformity, and focus accuracy must all be achieved at the nanometer level.
Second, computational imaging and algorithms. The raw images obtained from the hardware are insufficient; they must be supplemented with ultra-high-speed TDI (Time Delay Integration) detection, high-sensitivity automatic thresholding, and automatic defect classification (ADC) powered by deep learning to filter out noise that "looks like defects."
Third, the contradiction between sensitivity and throughput. Capturing smaller defects requires slower, brighter, and more sensitive processes; however, wafer fabs calculate costs based on wafers per hour (WPH). Every increase in sensitivity level multiplies the pressure on throughput and false alarm rates.
According to VLSI data, in 2023, bright-field nano-patterned wafer defect inspection equipment accounted for 19.5% of the global metrology and inspection market, representing the largest share among single categories. Only by developing such equipment and introducing it into mass production lines can one truly touch the upper limit of front-end process control.
| Global Landscape: KLA Dominates the Market
On a global scale, bright-field inspection, and indeed the entire front-end metrology and inspection sector, is a highly concentrated market.
KLA's market share varies depending on the statistical caliber, but the conclusion is consistent: it is the "sole superpower" in this market. According to disclosures from the company's 2026 Investor Day, its market share in semiconductor process control has reached 58%, an increase of 360 basis points compared to 2021. According to the caliber of Frost & Sullivan, the global metrology and inspection equipment market was approximately USD 17.44 billion in 2024, with KLA alone accounting for about 50%.
Following behind are Applied Materials, Hitachi High-Tech, as well as ASML (HMI) and Onto Innovation, which enter the market through two technological lines: electron beam and bright-field. The overseas CR3 (Concentration Ratio of the top 3 firms) has long exceeded 75%.
Market share is not built on price alone; three layers of barriers stand behind it:
Patent wall. Core patents for optical design, algorithms, and calibration methods are basically locked up by incumbent players, making it very difficult for latecomers to bypass them.
Validation cycle. Production line validation at leading wafer fabs generally takes 18 to 24 months. Equipment issues affect the yield of the entire line, giving customers no incentive to switch.
Installed base. The yield data generated by KLA equipment on advanced process nodes production lines globally is, in itself, a barrier.
There is another layer: under export controls, the supply of high-end inspection equipment related to advanced process nodes to China is inherently restricted. This is both a choke point and the only window for domestic companies. Customers are willing to provide validation opportunities, on the premise that you must deliver something close to being usable.
| What Domestic Companies Are Doing
In 2025, the Chinese semiconductor metrology and inspection equipment market was approximately USD 4.45 billion, with the localization rate of process control equipment at only about 4.2%, making it the weakest segment in terms of self-reliance among front-end equipment. Breakthroughs by domestic manufacturers basically follow the path of "from easy to difficult, from back-end to front-end, and from mature nodes to advanced nodes."
Skyverse: The first listed company in China for front-end optical metrology/inspection, with 2025 revenue of approximately CNY 2.053 billion. Products that have passed validation by leading domestic customers include bright-field equipment for mature process nodes. Regarding product lines under validation, the production line validation for bright-field nano-patterned wafer defect inspection equipment targeting advanced process nodes is progressing smoothly, and it is expected to see more positive results within the year; the validation of wafer flatness metrology equipment on the production lines of leading domestic HBM customers is progressing smoothly; and electron beam critical dimension metrology equipment has been shipped to leading customers' production lines for validation.
Shanghai Precision Measurement (Precision Measurement Semiconductor): Strong in the metrology end, including film thickness, OCD (Optical Critical Dimension), and overlay metrology. Its main film thickness and OCD systems have completed comprehensive delivery and acceptance for the 7nm advanced process node production lines at high-end wafer foundries. In the half-year from the end of 2025 to May 2026, it signed CNY 516 million in metrology/inspection contracts. Its depth lies in "metrology," while bright-field "inspection" is not its main battlefield.
It is evident that the common bottleneck for domestic manufacturers is exactly the bright-field inspection that Siok Technology is betting on, and it must reach advanced nodes. Skyverse's bright-field equipment for advanced process nodes is currently still undergoing production line validation; Precision Measurement's bright-field equipment has completed production line acceptance at the 28nm node and is advancing delivery/validation at the 14nm node. Neither company has currently completed mass production acceptance for 7nm bright-field inspection, leaving room for domestic substitution in this segment.
| Why Take the "65nm → 40nm → 14-28nm" Path
Siok Technology's product roadmap is very clear: the TB1100 corresponds to the 65nm node, the TB1500 advances to the 40nm node and has been delivered to customers for mass production, and the TB2000 targets the 14nm to 28nm nodes and is currently undergoing validation at leading fabs.
This path of climbing from mature nodes to advanced nodes is not conservative, but pragmatic in engineering. The core of inspection equipment consists of three parts: optical system design and manufacturing, image and algorithm processing, and whole-machine motion control and stability. These capabilities are continuously transferable across nodes. First, running through engineering, supply chain, and customer channels at the 40nm node to obtain real production line defect data and feedback, and then tackling the higher requirements for sensitivity and false alarm rates at the 14nm to 28nm nodes, is much more reliable than rushing directly to below 7nm in one step.
This also aligns with the team's background, with the core strength coming from industrial vision. Industrial vision is inherently about "using cameras to see things and using algorithms to find flaws," with a ready foundation in optical imaging, image algorithms, and high-speed motion control; the difference is that in the semiconductor front-end, this system must be elevated from "micrometer-level, allowing a certain rate of false alarms" to mass production standards of "nanometer-level, extremely low false alarm rate, and 24/7 non-stop operation." First proving at the 40nm level that the equipment can enter the fab, run stably, and help customers improve yield, and then attacking the 14nm to 28nm nodes, is the route that gives customers the confidence to provide validation opportunities.
The local semiconductor industry cluster in Suzhou, coupled with the entry of industry capital specializing in semiconductors such as Oriza Partners, also indicates that local governments and professional institutions recognize this step-by-step climbing route.
| Industry Perspectives
On the route: Not jumping hard from mature nodes to advanced nodes, but first running through engineering and customer channels before tackling the hard parts, is the most realistic approach for domestic inspection equipment. Siok Technology did not go head-to-head with KLA at the 3nm node, but instead found a foothold in the 14nm to 28nm segment, where domestic products are almost blank and customers have real demand, showing clear positioning.
On the environment: A 4.2% localization rate, an overseas CR3 exceeding 75%, and tightening export controls—stacking these three facts together leads to only one conclusion: as long as the product is close to being usable, wafer fabs have the motivation to validate and put domestic equipment on their production lines. The demand is rigid, which is the biggest support for companies like Siok Technology.
The track is advancing rapidly. Skyverse's REDWOOD-900 has already entered fabs in small batches, JingYuan Electron is flanking on the electron beam side, and Precision Measurement is deeply rooted in the metrology end. The domestic blank space in bright-field inspection is being filled by multiple companies simultaneously; whoever gets mass production acceptance first will set the benchmark for domestic bright-field inspection.
| Conclusion
Siok Technology's case, on the surface, is the contrast of "CNY 28,900 revenue and a CNY 2.2 billion valuation," but underneath, it is the process of domestic front-end inspection equipment companies collectively tackling the bright-field high ground. The difficult problem in bright-field inspection of extracting defects from pattern noise is the real capability that allows KLA to occupy half the market, and it is also the mountain that all domestic manufacturers must climb.
The validation of the TB2000 at leading fabs is a crucial step for Siok Technology moving forward; whether bright-field equipment can run through mass production acceptance from 0 to 1 domestically is the critical leap for the self-reliance of the entire front-end process control. Siok Technology has chosen a clear path, leaving the rest to the production lines to prove.
The financing, financial, and product roadmap data in this article come from company announcements and public reports; market size, share, and localization rates are cited from third-party research such as Frost & Sullivan, Soochow Securities, and Research and Markets. This article is for industrial observation only and does not constitute investment advice; semiconductor equipment validation cycles are long, progress is uncertain, and third-party data calibers vary, so it is for reference only.
Risk warning: This article is an industrial and financial analysis and does not constitute any investment advice. New share subscriptions carry risks, and entering the market requires caution.
Original article by Shishuo Xinyu, welcome to follow and share