Global Technology said it will accept subscriptions from retail investors for two days starting Sept. 16.

Global Technology is scheduled to list on the KOSDAQ on Sept. 29. Following five trading days of bookbuilding with institutional investors starting Sept. 7, the company finalized its offering price at 10,000 won on Sept. 15.

"Our fundamental technological capabilities and earnings growth potential remain unchanged," Global Technology said. The company has low-power, constant-current time-division driving technology that reduces the number of driver integrated circuits (ICs) required while maintaining the same image quality. Cumulative IC sales have exceeded 200 million units.

Global Technology is a fabless semiconductor company. It develops and supplies both ICs and modules. The company also mass-produces module-level products, including display modules and backlight units (BLUs) for automotive customers. Revenue reached 28.8 billion won in the first half of this year, approaching its full-year revenue of 29.7 billion won last year. The company also swung to a profit of 1.155 billion won.

The company said it has also minimized the potential overhang. Shares available for trading immediately after Global Technology's listing will account for 22.19% of its total shares. The remaining 77.81% will be subject to lock-up periods ranging from one month to as long as four years. That compares with 37% for fabless company Haechitech, which listed last month.

"We minimized the number of shares available for trading on the listing date to protect investors," Global Technology CEO Kim Min-sun said. "After the listing, we will drive corporate value by delivering tangible results through diversification of next-generation mobility products."

Rebellions said on Sept. 16 it will supply racks to Japanese artificial intelligence (AI) company ai&’s data center. The company plans to expand the supply to more than 100 units.

Rebellions will deploy its heterogeneous infrastructure system “REBELRack” at ai&’s data center. An initial batch will be installed at ai&’s Tokyo data center, with additional units to follow. Rebellions declined to disclose the volume or contract value, citing a nondisclosure agreement (NDA) with the customer.

Heterogeneous infrastructure combines different computing devices, including neural processing units (NPUs), graphics processing units (GPUs) and central processing units (CPUs). Rebellions will provide ai&’s data center with a heterogeneous infrastructure solution focused on power efficiency and lower operating costs.

REBELRack is Rebellions’ fully integrated rack-scale AI infrastructure. It supports the open-source frameworks used by ai&, allowing the system to be deployed without a large-scale infrastructure transition. Rebellions expects the system to begin operating at ai&’s data center on the day of delivery.

The system supports up to 32 Rebellions accelerator cards and eight AMD EPYC 9355 CPUs. It has 6 terabytes of DDR5 system memory and delivers 64 petaflops (PFLOPS) of performance using 8-bit floating-point (FP8) operations.

Rebellions said a configuration of 100 REBELRack units would have maximum power consumption of 2.8 megawatts (MW) for the rack equipment alone. The company did not disclose when delivery of all 100 units would be completed. ai& plans to build out its data center to a capacity of 40 MW by the end of 2027.

SK hynix workers approved a revised 2026 wage and collective bargaining agreement in a union vote, three weeks after the first tentative agreement was rejected by a margin of 25 votes. The deal was reached after raising the cash portion of performance bonuses to 50%.

The revised agreement was approved with 8,731 votes, or 57.08%, in a two-day vote held by SK hynix's Icheon and Cheongju union from Sept. 15, according to results released Sept. 16. Of 16,039 eligible members, 15,297 voted, for a turnout of 95.37%.

The basic payout ratio for the company's profit-sharing (PS) bonus was adjusted to 50% cash and 50% stock. The first agreement called for 40% in cash and 60% in stock. The revised deal also gives employees the option to convert the cash portion into stock in increments of 10 percentage points.

Deferred portions of performance bonuses calculated based on last year's results that had been scheduled for payment next year and the following year will also be paid earlier. Single-parent households were added to the eligibility for additional housing loan support, which previously covered married households. The broader framework of the previous agreement, including a 6.3% wage increase and the deferral of a portion of wages in the event of losses, was maintained.

The first tentative agreement was rejected in an all-employee union vote on Aug. 25, with 7,535 votes, or 50.08%, against and 7,510 votes, or 49.92%, in favor. Labor and management then held about two weeks of additional negotiations and reached the revised agreement on Sept. 9.

SK hynix and the union will hold a signing ceremony for the final agreement before the Chuseok holiday, completing this year's wage and collective bargaining process.

"We thank the labor union and our employees for working with us through this difficult process," SK hynix said. "We will continue to work together as a company and employees to address and overcome the challenges we face."

Samsung Electronics plans to outsource all additional production of conventional memory modules as it focuses its in-house capacity on high-value memory products such as high-bandwidth memory (HBM).

Samsung Electronics is strongly urging outsourced semiconductor assembly and test (OSAT) partners to expand production lines to meet growing demand for conventional memory modules, according to industry sources on Sept. 15. Instead of expanding its own production capacity for DDR5 memory modules and solid-state drives (SSDs) used in PCs, servers and laptops, the company is increasing allocations to outsourcing partners.

“Samsung Electronics lacks space to expand existing back-end production lines needed to manufacture semiconductors for artificial intelligence (AI), so it is reallocating space and equipment at packaging plants in Cheonan and Onyang to advanced packaging lines, including HBM,” a person familiar with the matter said. “The company has decided to handle most increases in conventional memory module production through outsourcing rather than expanding its own capacity.”

Another person familiar with the matter said, “Samsung Electronics has asked its partners since June last year to increase investment in production lines for conventional DDR5 memory modules and SSDs. This year, it has even proposed that partners bring forward capacity expansions from their existing plans.”

Samsung Electronics is expanding outsourcing of conventional modules because it needs to use its limited production facilities more efficiently. The company plans to begin construction this month of a new 6 trillion won HBM plant at its Onyang campus, but completion will take several years.

A new $1.5 billion back-end processing plant under construction in an industrial park in Thai Nguyen Province in northern Vietnam will also be dedicated to testing older memory products, including DDR4, low-power LPDDR4, NAND flash and universal flash storage (UFS) products. That leaves Samsung Electronics with limited in-house facilities available to increase conventional module production.

DDR5 memory modules are assembled by mounting prepackaged DRAM and NAND chips and passive components onto printed circuit boards (PCBs) using surface-mount technology (SMT). The process is relatively less complex than advanced packaging for products such as HBM, allowing outsourcing partners to expand production capacity sufficiently.

Hanyang Digitech, SFA Semicon and Dreamtech Expand Facilities

Samsung Electronics’ outsourcing partners are also expanding production facilities in line with the company’s increased outsourcing of memory modules.

Dreamtech, which began mass production of DDR5 modules in India in November last year, is also expanding its production lines. In June, the company approved equipment investment to convert and expand its Noida Plant 1 in India, which had assembled printed board assemblies (PBAs) for smartphones, into a large-scale memory module production line. The expansion work will be completed this month, followed by a gradual increase in production capacity. The plant is ultimately expected to secure annual production capacity of more than 50 million units.

Hanyang Digitech is working to improve productivity at its existing memory module plant in Vietnam. The company invested more than 31.5 billion won in the first half of this year to introduce new machinery and automate production processes.

SFA Semicon is transferring DDR5 testing and module assembly equipment previously used at Samsung Electronics’ Onyang campus to SFA Semicon’s plant in the Philippines. The first phase is scheduled to begin operations in November, followed by the second phase in the second quarter of next year. The move is expected to significantly increase DDR5 final testing and module production capacity. SFA Semicon handles fine-pitch ball grid array (FBGA) packaging of memory chips at its Cheonan plant in South Chungcheong Province and conducts final testing and module assembly at its Philippine plant.

Global Technologies has set its final initial public offering price below its proposed range ahead of its KOSDAQ listing.

Global Technologies said Sept. 15 that it set the final offering price at 10,000 won per share. The company had proposed a price range of 13,000 won to 15,000 won. The final price was as much as 33.3% below the proposed range. The offering will raise a total of 40 billion won, with the company expected to have a market capitalization of about 196.5 billion won.

Global Technologies said it decided on the price after taking into account "the subdued IPO market and increasingly cautious investor sentiment," adding that it sought to build medium- to long-term trust with investors rather than focus on short-term demand for the offering.

The company conducted bookbuilding for institutional investors over five trading days from Sept. 7 to Sept. 11. A total of 961 domestic and overseas institutions participated, resulting in a subscription ratio of 123.19 to 1.

Global Technologies designs light-emitting diode (LED) driver integrated circuits (ICs). While it operates as a fabless chip company, it has a pilot line for microLED-related products in Pyeongtaek. The company supplies not only ICs but also products in module form, including backlight units (BLUs). Its main customers are in the TV and monitor markets. It has recently been expanding into the automotive LED market.

Global Technologies passed the KOSDAQ preliminary listing review in July. It is pursuing a technology-special listing. Public subscriptions will be held Sept. 16-17, with the listing scheduled for Sept. 29. Proceeds from the listing will be used from this year through 2028 for wafer costs, research and development (R&D), and a microLED development line.

The company is targeting revenue of 169.5 billion won in 2029. Revenue in the first half of this year was 28.8 billion won, equivalent to 97% of its full-year revenue of 29.7 billion won last year. It also swung to a profit of 1.155 billion won during the period.

Qualcomm has signed a second agreement with a large-scale data center operator, or hyperscaler, to jointly design and produce artificial intelligence (AI) accelerators. The deal marks tangible progress in AI for Qualcomm, which has historically relied on smartphone chips for most of its revenue.

Akash Palkhiwala, Qualcomm’s chief financial officer (CFO), said at the Goldman Sachs Technology Conference 2026 on Sept. 8 that the company is working with another global hyperscaler in addition to Amazon. He said the custom silicon Qualcomm is developing with hyperscalers is intended not for auxiliary workloads but as primary AI accelerators in data centers.

The remarks indicate that Qualcomm is developing AI accelerators optimized for the customer’s data centers and services. Palkhiwala did not identify the hyperscaler. Major hyperscalers include Amazon Web Services (AWS), Microsoft, Google, Meta, Oracle and IBM. Qualcomm is already working with Amazon and Meta.

The agreement with the undisclosed hyperscaler is expected to be similar in structure to Qualcomm’s Amazon deal. Qualcomm and Amazon recently announced a collaboration on customized silicon for data center AI acceleration and high-performance optical connectivity solutions.

Qualcomm will supply the products to Amazon over the next 10 years. The agreement is valued at a total of $60 billion, or about 80.514 trillion won. Amazon will also receive warrants that vest in proportion to its purchases. The structure is intended to strengthen and extend the companies’ long-term partnership. Qualcomm has begun chip production, while Amazon has secured 15% of the warrants.

“The Amazon agreement is very large, but it is only the beginning,” Palkhiwala said. He added that Qualcomm’s technology collaboration with another hyperscaler is taking a similar form and is making strong progress.

Qualcomm has also agreed to supply data center central processing units (CPUs) to Meta. Its Dragonfly C1000 is planned for Meta’s next-generation server fleet. Dragonfly is Qualcomm’s data center product portfolio. Qualcomm is targeting more than twice the performance per watt of existing competing server CPUs with the C1000. A server fleet refers to a group of servers operating together to perform the same function.

Qualcomm expects its agreements with Amazon, Meta and the undisclosed hyperscaler to help push data center product revenue above 20 trillion won in three years. “We have already made significant progress toward our fiscal 2029 target of $15 billion, or about 20.07 trillion won,” Palkhiwala said. “We are at a point where we don’t need that many additional business elements.”

Qualcomm has recently been focusing its experience in advanced-node chip design and high-volume manufacturing on its custom silicon business. It is also developing Dragonfly commercial AI accelerators that are not customized for individual customers: the AI250 and AI300. The AI250 is scheduled for mid-2027, followed by the AI300 in 2028. Both products will incorporate Qualcomm High Bandwidth Compute (HBC) memory technology.

Global Technologies aims to generate 169.5 billion won in revenue by 2029 by expanding its light-emitting diode (LED) driver IC, transparent display, micro LED and automotive system semiconductor businesses.

Kim Min-sun, chief executive officer of Global Technologies, said at a press briefing at the CCMM Building in Yeouido, Seoul, on Sept. 10 that the company plans to expand its LED driver IC technology from TVs to monitors and automobiles, while developing micro LED and automotive system semiconductors as new growth engines.

Global Technologies is a fabless chipmaker that designs LED driver ICs (LDIs) controlling the current and brightness of backlight LEDs. The TV market currently accounts for about 70% of its revenue. The company secured a purchase order from a Taiwanese display customer in July. It plans to begin mass production of monitor LDIs in December and is developing automotive products with a target of mass production in the second half of next year.

The company is also developing a red-green-blue (RGB) backlight module combining driver ICs, LED packages and lenses. The key technology is time-division driving, which alternately controls three light sources through a single output channel. Conventional designs use a separate channel for each light source. By sharing the control circuitry, the technology can reduce the number of driver ICs required to as little as one-third of the conventional level.

Global Technologies plans to use the reduced chip count to counter price competition from Chinese suppliers. TV makers can lower overall component costs without cutting the price of individual chips. The company aims to improve profitability by charging higher prices for chips incorporating the technology while increasing shipment volumes as customers expand their adoption of RGB LED models. A TV-set customer will evaluate the technology this month, with mass production targeted for early next year.

In transparent LED displays, the company is targeting commercial signage and automotive glass. The products combine driver ICs with ultra-small LED packages, allowing users to see through the display. Global Technologies has supplied commercial products to customers including Yangju City Hall in Gyeonggi Province and is developing products for vehicle rear windows and roofs.

In micro LEDs, the company is jointly developing pixel driver ICs, packages and modules. A 24-pixel driver IC for TVs is targeted for mass production early next year. The company is also developing modules for automotive center information displays (CIDs). It invested 5 billion won to establish a pilot line in Pyeongtaek for prototype production and process validation.

Global Technologies is developing lighting and communications products for its automotive system semiconductor business. It operates a joint research center with an automotive parts supplier and is developing about 10 products. Its smart ambient lighting (SAL) IC, which controls vehicle interior lighting, has secured an initial order for 3 million units, with deliveries scheduled to begin early next year. The company also plans to commercialize controller area network (CAN) communication chips used to exchange data among vehicle electronic systems.

The company posted first-half revenue of 28.8 billion won and operating profit of 1.155 billion won this year. It generated revenue equivalent to 97% of last year's full-year total in the first half and returned to an operating profit. Global Technologies plans to increase revenue from mass-produced products and raise its operating margin from a target of 7.7% this year to 21.8% in 2029.

Global Technologies is scheduled to list on the KOSDAQ on Sept. 29. It is offering 4 million shares at an indicative price range of 13,000 won to 15,000 won each, for an offering size of 52 billion won to 60 billion won. The company is conducting bookbuilding for institutional investors through Sept. 11 and will accept retail subscriptions on Sept. 16 and 17. Korea Investment & Securities is the lead underwriter.

The company plans to invest a combined 78.7 billion won through 2029 using IPO proceeds, operating cash flow and government funding. Of the total, 23.5 billion won will be allocated to research and development, 7.7 billion won to facility expansion and 47.5 billion won to working capital, including securing foundry capacity.

SK hynix plans to save a cumulative 600 million tons of water at its Korean operations by 2030.

The company said it is presenting its water conservation achievements and targets at Korea International Water Week 2026, held at EXCO in Daegu from Sept. 9 to 11. Its exhibition booth highlights major water conservation, recycling and ecosystem preservation activities through videos and graphics.

SK hynix has operated a water and wastewater reduction task force since 2018, reducing water consumption and expanding reuse in semiconductor manufacturing. As a result, its Korean operations saved an average 170,000 tons of water per day last year, equivalent to the daily household water consumption of about 550,000 people.

Cumulative water savings reached 303.37 million tons by the end of last year, exceeding the company’s target of 264 million tons. SK hynix aims to increase the total to 329 million tons this year. Water reuse rose about 54% to 73.59 million tons last year from 47.88 million tons in 2022.

SK hynix’s Icheon site produces 94,400 tons of recycled water per day through its wastewater reuse facilities. The recycled water is supplied to facilities including air pollution control systems. Its Cheongju site became the first in Korea’s semiconductor industry to introduce externally treated reclaimed wastewater in 2023, recycling it for industrial use.

The company is also strengthening water management across the entire cycle from intake and use to recycling and discharge. Its wastewater treatment processes include ozone oxidation, activated carbon and multistage coagulation and sedimentation systems. SK hynix continues to develop and invest in additional technologies to reduce pollutants.

SK hynix also carries out activities to monitor local communities and ecosystems. It operates the ECOSEE citizen science program around Anseong Stream to track ecological changes before and after development of the Yongin semiconductor cluster. Citizens participate directly in surveys of plants, birds and aquatic organisms and in collecting data.

“We will continue to advance our water management system to use less water, which is essential to semiconductor production, and reuse as much of it as possible,” SK hynix said. “We will also steadily expand investments in process improvements and related technologies and facilities.”

SK hynix management and its labor union reached a revised tentative agreement on Sept. 9 that sets the basic performance bonus payment ratio at 50% cash and 50% stock. The union will hold a temporary delegates' meeting on Sept. 10, followed by a membership vote from Sept. 15 to 16.

The revised agreement retains the broad framework of a tentative deal reached last month while reducing the proportion of bonuses paid in stock. The previous agreement called for profit-sharing (PS) bonuses to be paid 40% in cash and 60% in stock. The revised plan sets both at 50%.

Employees can choose to increase the stock portion in increments of 10 percentage points, up to 100%.

Deferred portions of performance bonuses calculated based on last year's results, which had been scheduled for payment next year and the following year, will also be paid earlier. The measure is intended to reduce confusion as the bonus system shifts from primarily cash payments to a combination of cash and stock.

The company also revised part of its housing loan support program. Additional support beyond the basic program had been limited to married households, but the revised proposal extends eligibility to single-parent families.

SK hynix's Icheon and Cheongju full-time employee unions previously held a membership vote last month on a tentative agreement that included a 6.3% wage increase and changes to the PS payment structure. The proposal was rejected by 7,535 votes, or 50.08%, against and 7,510 votes, or 49.92%, in favor — a difference of just 25 votes.

Startup Zeta Mobility is expanding its 4D radar artificial intelligence sensor-fusion technology, originally developed for autonomous driving, into senior care and healthcare. The company plans to produce tangible results this year through partnerships with a company and a university.

“We signed memorandums of understanding with two companies and institutions last month and began proof-of-concept testing,” Zeta Mobility Vice President Kim Sang-ki said on Sept. 9. The partners are senior care technology company Dolbomdream and Gachon University’s Intelligent Homecare Integrated Support Infrastructure Project Group, or AAL Center. Zeta Mobility aims to commercialize services for vulnerable populations using its technology.

Zeta Mobility is a deep-tech startup developing high-resolution 4D imaging radar and AI sensor-fusion technologies. The company was founded in February 2024 as a startup spun out of a laboratory at KAIST’s Graduate School of Mobility and joined the university’s startup incubator. In October that year, it was selected for the Ministry of SMEs and Startups’ Deep Tech TIPS program and Samsung Electronics’ C-Lab Outside startup support program. It was also selected for the ministry’s Super Gap Startup 1000+ program in April last year and received an award from the head of the Small and Medium Business Administration in May.

Radar emits electromagnetic waves, or millimeter waves, and analyzes reflected signals. Unlike conventional 3D radar, which measures distance, direction and speed, 4D imaging radar also captures height data. This allows scattered radar reflections to be converted into high-resolution three-dimensional point-cloud data.

Raw 4D radar data contains signal noise that makes it difficult to interpret. Zeta Mobility addresses this with its proprietary AI algorithm software, Zeta AI Engine. The software removes meaningless reflections and signal noise and extracts the shape of objects. By analyzing continuous time-series data, it can infer human movement, changes in posture and biometric signals such as heart and respiratory rates with more than 90% accuracy.

Zeta Mobility has integrated the technologies into three versions of its ARGUS platform, an intelligent system that detects, understands and transmits radar sensor data. The platform takes the form of a compact, low-power, high-efficiency board for edge AI, which runs AI models directly on devices where data is generated.

ARGUS-A is designed to detect safety conditions and driving environments for autonomous vehicles. ARGUS-C monitors vehicle occupants in real time. ARGUS-H is an indoor care product designed to detect falls among elderly people living alone and monitor biometric signals.

Zeta Mobility initially focused on autonomous driving and in-cabin occupant-detection sensors. But as commercialization and market adoption of autonomous driving progressed more slowly than expected, the company strategically shifted its business focus toward senior healthcare. Its 4D radar AI and sensor-fusion technologies can be applied to indoor residential spaces as well as vehicle interiors.

Partnerships With Dolbomdream, Gachon University to Test ARGUS-H

Zeta Mobility has secured infrastructure to test ARGUS-H through the two recently signed MOUs.

The company will first work with Dolbomdream on joint testing and commercialization of senior care and monitoring services. The companies plan to combine Zeta Mobility’s radar AI sensor technology with Dolbomdream’s wearable sensor-based healthcare technology and jointly pursue government projects and research and development programs. They also plan to expand their collaboration into worker safety management at industrial sites.

Existing senior-care technologies face trade-offs between accuracy, convenience and privacy. Falls among elderly people living alone frequently occur in private spaces such as bathrooms, toilets and bedrooms, where cameras face strong user resistance as well as ethical and legal constraints. Smartwatches and patch-type biosensors require users to wear them continuously and recharge them regularly. Elderly users may forget to wear the devices or take them off, creating monitoring gaps and potentially delaying responses to emergencies such as falls.

Zeta Mobility says its 4D radar AI technology can address those problems. ARGUS-H can be installed on a wall or ceiling to monitor an entire space without requiring the user to wear a device. Because it analyzes radar waveforms rather than visual images, it avoids capturing private visual information. It can also detect a person’s precise location and subtle movements in low-light or steam-filled environments where visibility is poor.

Zeta Mobility will work with Gachon University’s AAL Center to develop and commercialize intelligent homecare services for vulnerable populations. The center has researched a big-data platform for elderly people’s activities of daily living. Zeta Mobility plans to combine the university’s software and data research capabilities with its own technology to build infrastructure for real-time monitoring of elderly users.

The partners will also develop and deploy AI care equipment and personalized service models. They plan to support healthcare programs for elderly people run by local governments and community welfare organizations, while expanding into the senior healthcare and smart-home markets.

“We will use these recent demonstration agreements as a stepping stone to establish technical standards for smart-home senior care and secure our first supply contract with a construction company or healthcare platform within this year,” Kim said.

SK keyfoundry said Sept. 9 it signed a long-term agreement to provide 130-nanometer foundry services to automotive semiconductor company Elmos Semiconductor.

SK keyfoundry recently held a signing ceremony in Frankfurt am Main, Germany, and will provide foundry capacity to Elmos through 2037. Elmos has used SK keyfoundry’s 350-nanometer process technology for the past 18 years.

“Elmos is an outstanding company in the automotive semiconductor market,” an SK keyfoundry official said. “Securing a long-term supply agreement is meaningful not only in terms of revenue but also for our technological growth.”

Wafer supply prices may be adjusted under the agreement. The 200-millimeter wafers supplied by SK keyfoundry are experiencing a global shortage as major chipmakers including TSMC and Samsung Electronics reduce 200-millimeter capacity and shift toward 300-millimeter production.

SK keyfoundry and Elmos will jointly develop the E550.01 Smart 4-Channel eFuse Controller. The chip controls the supply of power inside a vehicle across four separate paths and cuts power to protect electrical systems when abnormalities such as overcurrent occur. Mass production is targeted for 2027.

SK keyfoundry will also provide Elmos with deep trench isolation (DTI) capability. The company said the technology enables highly integrated, high-voltage designs with low substrate current.

“SK keyfoundry has been a trusted partner for many years,” Elmos Chief Executive Officer Arne Schneider said. “It will support our continued growth.”

“We will accelerate our automotive semiconductor supply, starting with Elmos,” SK keyfoundry CEO Lee Dong-jae said.

DB HiTek is in the final stages of preparing for mass production of its 8-inch silicon carbide (SiC) wafer foundry business.

DB HiTek said Sept. 9 that it had completed reliability verification of a 1,200-volt SiC MOSFET. The company plans to provide customers with its third-generation process design kit (PDK) in November as it moves to launch its foundry business.

The third-generation PDK targets specific on-resistance of 2.3 mΩ·cm² at an 18-volt gate voltage, down from 2.5 mΩ·cm² for the second generation. It is also designed to allow the device to withstand a sudden short circuit for 2.5 microseconds without damage. DB HiTek said the third-generation PDK could shorten semiconductor design time by more than a year.

“We have secured the core technology and mass-production foundation for what will be the world’s first 8-inch SiC foundry service,” DB HiTek said. “We will prepare for mass production as planned and move to establish an early lead in the market.”

Mass production is targeted for the first half of next year. Rival SK keyfoundry is also expected to begin mass production of 8-inch SiC foundry products in the first half of next year. SiC foundry operations require production lines to be developed and established jointly with customers, meaning both companies have effectively secured customers ahead of mass production. Samsung Electronics also has an 8-inch SiC production line, but does not offer it to external foundry customers.

The SiC semiconductor market currently centers on 150-millimeter wafers, while mass production using 200-millimeter wafers remains rare globally. SiC semiconductors use specially processed wafers that differ from conventional silicon (Si) wafers. Due to the technology’s lower maturity, production yields for 200-millimeter SiC wafers are lower than for 150-millimeter wafers, resulting in lower output.

DB HiTek will operate its SiC foundry at its Sangwoo Campus in Eumseong, North Chungcheong Province, converting the current development line into a mass-production line. Initial capacity will be relatively small at 800 wafers a month, about 1.3% of the Sangwoo facility’s total monthly capacity of 63,000 wafers. The company has established an integrated process covering design, manufacturing, characterization, reliability verification and epiwafer supply.

For now, DB HiTek is operating the development line primarily for customers it has already secured. It plans to add customers after establishing the mass-production system and gradually expand capacity in line with market demand.

Qualcomm has announced a semiconductor partnership with Amazon for artificial intelligence data centers, tied to warrants and a potential deal worth up to $60 billion.

Qualcomm said Sept. 8 it began a product collaboration with Amazon to build next-generation AI data center infrastructure. The partnership aims to supply cost-efficient custom application-specific integrated circuits (ASICs) for large-scale AI inference workloads on Amazon Web Services (AWS). ASICs are dedicated chips optimized for specific computing tasks or services.

"As AI demand accelerates, data center infrastructure will require advances in both computing and connectivity to deliver greater performance with more efficiency," Qualcomm Chief Executive Officer Cristiano Amon said. "Qualcomm is pleased to work with AWS on customized silicon and connectivity solutions, bringing decades of leadership in advanced processing and power-efficient compute, to deliver breakthrough performance and enable the next generation of AI infrastructure."

The companies will also jointly design high-performance optical connectivity solutions to address rising data traffic and bandwidth requirements in data centers. The solutions will target transmission speeds of up to 1.6 terabits per second (Tbps), using Qualcomm's high-speed serializer/deserializer (SerDes) and optical digital signal processor (DSP) technologies.

SerDes chips convert slower parallel signals inside semiconductors into a single high-speed serial data stream for transmission. Optical DSPs precisely compensate for signal distortion and loss as optical data travels through fiber.

Qualcomm Chief Financial Officer Akash Palkhiwala said the Amazon collaboration could generate a potential revenue opportunity of as much as $60 billion. He added that Amazon's selection validates Qualcomm's data center technology, strategy and ability to supply products at scale.

Up to $60 Billion Over 10 Years

The potential revenue cited by Palkhiwala is linked to Qualcomm's issuance of warrants, which give Amazon the right to purchase shares at a specified price in the future. Qualcomm plans to vest the warrants in stages based on Amazon's purchases of its data center ASICs and optical connectivity solutions. Amazon's maximum contracted purchase amount is $60 billion, with the agreement running for 10 years.

Amazon has already secured 15% of the warrants after committing to purchase Qualcomm products at the initial stage of the agreement. The product volume is estimated at about $9 billion.

Qualcomm has also received purchase orders (POs) from Amazon and begun chip production. Revenue from the chips is expected to be recognized starting in December. That makes Qualcomm's previously announced fiscal 2027 data center revenue forecast of about $5 billion increasingly likely to be achieved.

Qualcomm also plans to expand its use of Amazon's AI platform to shorten semiconductor design cycles. The company will increase the use of services including Amazon Bedrock in electronic design automation (EDA) workloads.

EDA involves using software to design semiconductor circuits and verify their performance. Amazon Bedrock is a fully managed generative AI service that allows developers to use pretrained generative AI models in cloud environments for their specific workflows and applications.

Qualcomm's data center ASICs are part of its Dragonfly portfolio. In addition to custom silicon, Dragonfly includes the C1000 central processing unit (CPU), AI300 AI inference accelerator and connectivity products.

Meta and Panmnesia have proposed a next-generation artificial intelligence (AI) data center architecture designed to make an entire data center operate like a single chip. The approach could address latency problems that emerge as AI models grow larger.

Panmnesia said Sept. 8 that it and Meta introduced technology in a Nature Reviews journal that connects central processing units (CPUs), accelerators and memory through Compute Express Link (CXL) and scales the architecture across a data center. Researchers from Meta and Panmnesia co-authored the paper.

The approach uses CXL to reduce variations in latency between racks, allowing a data center to operate predictably like a single chip. CXL is a high-speed interface that connects different devices including CPUs, graphics processing units (GPUs) and memory.

Racks, the building blocks of AI infrastructure, exchange data rapidly within each rack using high-speed link technologies. Standard interfaces such as Ethernet and InfiniBand are used to connect one rack to another.

The problem is that as AI models expand to trillions of parameters, latency with conventional interfaces increases sharply. AI companies have increased the number of CPUs and GPUs to train on massive datasets, inevitably resulting in higher latency.

Meta and Panmnesia addressed latency variation by connecting racks through CXL. They added three types of hardware: high-fanout switches, fabric controllers and link acceleration units. They also proposed optical CXL to reduce the physical distance over which signals must travel.

The number of accelerators that a single CPU can manage increased from two to 16, while up to 960 connected domains can be managed. Latency was reduced by as much as 90%. Panmnesia said the CXL-based technology demonstrated that models larger than today's AI models can be trained without interruption.

"CXL presents a direction for next-generation AI infrastructure in which a data center operates as a single computing system," Panmnesia CEO Jung Myoungsoo said.

Samsung Electronics said Sept. 9 that it will jointly develop artificial intelligence (AI) models with Mistral AI specialized for semiconductor design and manufacturing.

The two companies formed a strategic partnership on the occasion of the South Korea-France summit held the same day. Samsung previously made a large equity investment in Mistral AI, laying the foundation for long-term technology cooperation and joint business initiatives. The Series D funding round for Mistral AI led by Samsung totaled €3 billion.

The companies will develop semiconductor-specific AI combining semiconductor technology and manufacturing data from Samsung’s Device Solutions (DS) division with Mistral AI’s AI model technology. The specialized AI will be deployed in stages across Samsung’s semiconductor operations, from chip design to manufacturing data analysis, defect prediction and process optimization. The goal is to shorten development time, improve manufacturing efficiency and quality, and enhance productivity and precision across semiconductor operations.

The companies will also jointly build an AI platform and operating system optimized for semiconductor operations using an on-premises approach. On-premises systems are built and operated on a company’s own servers and data centers. The approach is intended to enable Samsung to securely use semiconductor-related data without transferring it outside the company.

Samsung plans to expand its cooperation with Mistral AI across its DS division, from memory to semiconductor foundry and logic businesses. It will also accelerate AI transformation (AX) across all business units. The company ultimately plans to build an AI-based semiconductor ecosystem that can connect customers and partners.

“As the complexity of semiconductor design and manufacturing increases rapidly, the ability to use vast amounts of data securely and precisely through AI is emerging as a key competitive advantage,” said Jun Young-hyun, vice chairman and head of Samsung Electronics’ DS division. “By building AI models and platforms specialized for the semiconductor industry with Mistral AI, we will establish a new standard for AI-driven semiconductor innovation.”

“AI is redefining the creation of advanced technologies, from semiconductors to software,” said Arthur Mensch, co-founder and chief executive officer of Mistral AI. “Through our collaboration with Samsung Electronics, we will contribute to advances in semiconductor design and manufacturing and across the global AI value chain.”

Mistral AI is a French AI company that has developed high-efficiency large language models (LLMs) including Mistral Large. It was founded by AI researchers formerly at Google DeepMind and Meta. The company has technology for building customized on-premises AI models for industries requiring high levels of data security, including finance, manufacturing and energy.

Arm is preparing for the spread of agentic artificial intelligence (AI) across mobile devices, data centers and robotics.

Arm unveiled “CSS for Mobile 2,” its latest compute subsystem (CSS) for mobile devices, on Sept. 7. CSS for Mobile 2 combines the Mali G2 Ultra NX graphics processing unit (GPU) with the C2 CPU Cluster.

The platform is optimized for edge environments as agentic AI evolves. Arm expects agentic AI workloads to increasingly shift from data centers to edge devices such as desktops, laptops and mobile devices, driven by high token costs, long latency and security concerns. The company expects on-device computing, which operates physically closer to users, to become increasingly important in addressing these issues.

The Mali G2 Ultra NX features a dedicated neural accelerator. It supports neural super sampling, which renders graphics at a lower resolution and then uses AI to reconstruct them at a higher resolution, extending technology previously used on desktops to mobile environments. Arm placed the neural accelerator close to the GPU to enable neural graphics at about 1 watt of power on mobile devices. Frame rates can increase by as much as fourfold at the same power level.

The C2 CPU Cluster enhances Scalable Matrix Extension 2 (SME2) capabilities for AI computing. While the previous C1 supported one SME2 unit, the C2 can be configured with up to two. SME2 performance in the C2 reaches 5 to 7 trillion operations per second (TOPS). General CPU performance has also improved, with single-thread performance up 15% and multi-thread performance up 12% from the previous generation. Processing time for agentic AI workloads is reduced by as much as 24%.

Arm also introduced “Neoverse CSS N4” for data centers. The platform is designed to accelerate development of custom systems-on-chip (SoCs) for general-purpose computing, networking and storage. It supports enhanced security, higher memory bandwidth, chiplet interfaces, LPDDR6 and PCIe Gen7. Arm increased design flexibility so customers can configure core counts, cache capacity and input/output (I/O) according to their requirements. Performance per socket is more than double that of the previous generation, while performance per watt improves 25% and memory bandwidth increases 75%.

Arm also provided an update on its first silicon chip, the “AGI CPU,” announced in March. The company said OpenAI, Meta, Cloudflare, Oracle, SAP and Lenovo are developing AGI CPU-based solutions following its launch.

Related products are expected to be released in the coming months. Customers that design their own SoCs can receive the technology as intellectual property (IP), while those seeking to rapidly expand computing infrastructure can choose finished products.

Arm is also strengthening its push into the physical AI market, including autonomous driving, robots and drones. The company estimates the physical AI computing market is currently worth $25 billion and expects it to grow to more than $200 billion after 2030.

Arm also announced the launch of Total Design for Physical AI, with more than 80 key companies including AWS, Hugging Face, NXP, Siemens and Unitree participating. The initiative links the entire development cycle — including silicon, cloud, sensing, security and system validation — with the physical AI ecosystem to support product development. Arm is also developing a common framework for robotics platforms, aiming to rigorously standardize robot capabilities by level, similar to the Level 1-to-5 system used for autonomous driving.

Synopsys is expanding its semiconductor design platform following its acquisition of Ansys, targeting further growth in South Korea’s semiconductor market.

Shankar Krishnamoorthy, chief product development officer (CPDO) at Synopsys, told a press conference at the Grand InterContinental Seoul Parnas in Gangnam on Sept. 8 that the company will “accelerate technological innovation in South Korea through agentic artificial intelligence (AI), multiphysics-based design and a silicon-to-systems integrated engineering approach.”

The importance of the South Korean market is growing. According to Synopsys’ own research, semiconductors are expected to account for 42.3% of the country’s total exports this year, while high-bandwidth memory (HBM) exports are projected to increase 153%.

Most of Synopsys’ revenue in South Korea comes from the memory sector, with Samsung Electronics and SK hynix among its major customers. The two companies use Synopsys’ semiconductor design tools and other technologies to develop memory chips.

Synopsys is also expanding its revenue base into system semiconductors. “Beyond memory, we work with Samsung Electronics’ foundry and System LSI businesses,” Krishnamoorthy said. “We also collaborate with Rebellions and FuriosaAI across all areas, including electronic design automation (EDA) tools.”

Synopsys’ strength lies in the breadth of its design environment. The company acquired engineering simulation software company Ansys in July last year, expanding beyond its existing EDA and intellectual property (IP) portfolio to add system design and simulation capabilities. This allows chips to be designed with consideration for the systems in which they will operate, including rockets, aircraft and data centers.

The company is also addressing the growing complexity of AI chip design. As AI chips incorporate more dies and HBM, interconnect density increases, requiring designers to consider signal integrity, power, thermal effects and stress simultaneously. Synopsys combines EDA and multiphysics technologies to support advanced designs including three-dimensional integrated circuits (3D ICs) and co-packaged optics (CPO).

Synopsys is also applying agentic AI to EDA. The technology assists engineers and autonomously performs lengthy tasks. In an agentic AI design verification process with Nvidia, the time required to obtain register-transfer level (RTL) results was reduced by a factor of 50, while verification coverage increased 20%. Productivity in analog and mixed-signal design improved by as much as threefold.

Synopsys also supports its own application programming interface (API). The company said it requires fewer tokens and lower token costs than competing solutions while significantly reducing latency.

SEMIFIVE has secured its first mass-production project using a 4-nanometer process. The company did not disclose the size of the supply deal, but said it exceeds 10% of last year's revenue of about 121 billion won.

SEMIFIVE said Sept. 8 that it will begin mass production of a 4nm chip for South Korean fabless semiconductor company HyperAccel. Production has been delayed by two quarters from the original schedule. In 2024, SEMIFIVE said it would design HyperAccel's "Bertha" using Samsung Electronics' 4nm process, with mass production initially targeted for the first quarter of 2026.

The company did not disclose the contract value due to a non-disclosure agreement (NDA) with the customer. The deal exceeded 10% of last year's revenue, making it subject to regulatory disclosure. The contract runs through June 15, 2028. Revenue will be recognized in stages as products are supplied, with a 10% upfront payment, 60% interim payment and 30% final payment.

Bertha is specialized for artificial intelligence (AI) inference workloads, including large language models (LLMs). The chip has a die area of more than 500 square millimeters. Naver Cloud is reportedly a potential customer.

As a design house, SEMIFIVE designed Bertha for optimization with Samsung Electronics' semiconductor foundry process. It handled front-end design, verification, packaging, testing and mass production on a turnkey basis. Under the contract, SEMIFIVE generates revenue each time the jointly designed chip is produced.

SEMIFIVE's mass-production orders totaled about 42.3 billion won in the first half of this year, roughly double the 21.2 billion won recorded last year.

"We will continue the growth in mass-production revenue demonstrated in the first half into the second half," SEMIFIVE CEO Cho Myung-hyun said. The company expects revenue to exceed 200 billion won this year. Company-wide revenue totaled about 94.7 billion won in the first half.

Samsung Electronics plans to become the first in the industry to introduce high-numerical-aperture (High-NA) extreme ultraviolet (EUV) lithography equipment from Dutch semiconductor equipment maker ASML into advanced DRAM manufacturing facilities by 2028. The companies will also jointly develop large 300mm photomasks.

Samsung Electronics announced the expanded cooperation with ASML on next-generation semiconductor manufacturing on Sept. 8.

"Building on the results of our existing cooperation with ASML, we will accelerate the development and adoption of next-generation technologies needed to manufacture advanced artificial intelligence (AI) semiconductors," Samsung said. "The cooperation means High-NA EUV technology is ready for full-scale application in advanced memory, following its use in foundry manufacturing processes."

High-NA refers to a numerical aperture of 0.55, a measure of a lens's ability to collect light. Low-NA EUV equipment has a numerical aperture of 0.33. The improved light-gathering capability allows High-NA systems to print finer and more precise semiconductor circuit patterns than Low-NA equipment.

Samsung plans to use the ultra-fine patterning technology to overcome the limits of DRAM scaling. It also aims to simplify circuit-forming processes and improve productivity.

○ Samsung Joins 300mm Photomask Consortium

Samsung will also participate in a large-mask consortium led by ASML. The consortium was formed to transition the photomask standard for semiconductor lithography equipment to 300mm.

Samsung will play a key role in the transition to 300mm photomasks based on its experience in memory manufacturing and semiconductor foundry production, as well as its track record in EUV operations. Consortium members will jointly review global technology trends in lithography equipment and photomasks.

A photomask is a precision template used to form microscopic electronic circuits on semiconductor wafers. Inside lithography equipment, light is projected through the mask to transfer circuit designs onto a wafer. Patterning precision affects semiconductor computing performance and yield.

The industry has traditionally used 150mm photomasks. With 150mm masks, High-NA EUV equipment has about half the exposure field of Low-NA EUV systems, meaning the area that can be patterned in a single exposure is smaller. Large chip circuits therefore need to be exposed in two sections. This is because the optical system was modified to prevent light interference and shadowing effects.

The separately exposed circuit sections must then be connected through a process known as stitching. Switching to 300mm masks eliminates the stitching constraint. This can reduce process steps, improve productivity and lower chip manufacturing costs. Samsung believes 300mm masks will be necessary to maximize the performance of High-NA EUV equipment in advanced semiconductor manufacturing, where tens of billions of transistors are densely packed onto a chip.

"Samsung Electronics is strengthening its leadership in advanced semiconductor manufacturing, including foundry and memory," said Jun Young-hyun, vice chairman of Samsung Electronics. "By strengthening our cooperation with ASML, we will establish the technology foundation for the next generation of AI."

"Samsung Electronics has long been one of ASML's most important partners, and we have advanced semiconductor manufacturing technology together," said Christophe Fouquet, chief executive officer of ASML. "In the new era driven by AI, we will continue to lead next-generation semiconductor manufacturing innovation together with Samsung Electronics."

The Korea Electronics Packaging and Circuits Association (KPCA) said Sept. 7 that it conducted an advanced semiconductor packaging industry training program in Taiwan.

The Taiwan advanced packaging program was jointly organized by KPCA, Gyeongsang National University and Pukyong National University. A total of 16 people participated, including students specializing in semiconductor packaging, two professors from Gyeongsang National University and two KPCA officials.

The program was designed to give Korean students firsthand experience of the global semiconductor industry and help them understand Taiwan's semiconductor packaging and printed circuit board (PCB) ecosystem. Participants visited copper-clad laminate maker Elite Material Co. (EMC), Lunghwa University of Science and Technology, the Taiwan Printed Circuit Association (TPCA) and SEMICON Taiwan 2026.

EMC is a materials company that produces copper-clad laminates (CCLs), which are used to manufacture PCBs. The visit gave participants an opportunity to observe semiconductor and electronics technologies and production sites. It also helped deepen their understanding of changes in the global supply chain and the direction of the advanced packaging industry.

At Lunghwa University of Science and Technology, participants examined Taiwan's industry-academia cooperation and semiconductor workforce training system. During their visit to TPCA, they learned about Taiwan's PCB and semiconductor packaging industries, global competitiveness, industrial ecosystem and workforce development policies.

At the final stop, SEMICON Taiwan 2026, participants explored the latest technologies and market trends across the global semiconductor industry. They reviewed developments in advanced packaging, high-bandwidth memory (HBM), chiplets and 2.5D and 3D packaging, technologies that are gaining importance amid the rapid growth of artificial intelligence (AI) semiconductors. SEMICON Taiwan is an exhibition for sharing advanced semiconductor technologies.

"For Korea to continue growing as a global semiconductor powerhouse, not only industry but also universities and research institutions need to work together to systematically develop future professionals," KPCA said. "KPCA will expand cooperation with universities, companies and related organizations to cultivate talented professionals who will lead the PCB and semiconductor packaging industries and strengthen global industry-academia collaboration."

KPCA is building a cooperative framework connecting industry, government, universities and research institutions. The effort is aimed at strengthening the competitiveness of Korea's PCB and semiconductor packaging industries and expanding global cooperation.