Attopsemi, a pioneering provider of innovative One-Time Programmable (OTP) IP solutions, today announced a significant technological leap: its proprietary I-fuse technology has achieved silicon validation on the 7nm FinFET process. Following the previous success on 12nm process, this latest milestone reinforces Attopsemi's technical agility in scaling its proprietary solutions alongside the cutting edge of semiconductors.Unlike conventional OTP solutions, I-fuse offers a comprehensive portfolio spanning multiple generations—from polysilicon to High-K Metal Gate (HKMG) and FinFET—without requiring additional mask layers or Boolean masking operations. This ensures seamless integration and high portability across a broad spectrum of process technologies.Attopsemi's revolutionary I-fuse is programmed below "thermal runaway." This allows electromigration (EM) to serve as the one and only one programming mechanism, ensuring high reliability guaranteed by fundamental physics. By utilizing metal as the fuse material, I-fuse benefits from the inherent scalability of Moore's Law: as metal width and height shrink, so do programming current and overall IP footprint. This scalability makes metal I-fuse stand out among other OTP technologies.The integration of I-fuse into advanced FinFET nodes enables a new generation of applications where small area, high reliability, low-power programming and low current read are critical. This OTP technology is poised to significantly impact the global semiconductor community, especially for AI and chiplets that need chips on very advanced nodes and from heterogeneous process technologies on the same package."Being able to scale I-fuse across FinFET nodes from 12nm to 7nm demonstrates I-fuse can be ported directly to 2nm GAA and beyond. The fundamental I-fuse programming mechanism for GAA is still the same, different MOS structures built by GAA are just to deliver currents." said Shine Chung, Chairman of Attopsemi.Attopsemi continues to make significant strides in providing high-performance, cost-effective OTP technologies. Its rapid growth and technical milestones demonstrate a ready-to-deploy solution now available across an expanding range of world-class foundries.

SK hynix Inc. (or "the company", www.skhynix.com) announced today that it has achieved record financial results in 2025 – 97.1467 trillion won in revenue, 47.2063 trillion won in operating profit (with an operating margin of 49%), and 42.9479 trillion won in net profit (with a net margin of 44%).The results significantly exceeded the previous record set in 2024. The annual revenue increased by more than 30 trillion won while the annual operating profit nearly doubled year-on-year, marking the highest annual performance in the company's history.Growth momentum accelerated further in the fourth quarter. In addition to HBM, demand on conventional memory solutions for servers increased sharply, to which SK hynix responded proactively. As a result, the company achieved record-high quarterly performance across all three indicators, with revenue rising 34% to 32.8267 trillion won, operating profit surging 68% to 19.1696 trillion won, and operating margin reaching 58% quarter-on-quarter.SK hynix emphasized that 2025 marked a year in which the company once again demonstrated its world-class technological leadership. In response to an AI-centric demand structure, the company secured both growth and profitability by enhancing technological competitiveness and expanding its portfolio of high value-added products.In the DRAM segment, HBM revenue more than doubled year-on-year, making a significant contribution to the company's record performance. Conventional DRAM entered full-scale mass production of 1cnm process, or the sixth-generation of the 10-nanometer technology. The company also solidified its leadership in server modules, with the development of the 256GB DDR5 RDIMM - a high capacity server module based on 32Gb fifth-generation 10nm-class (1b) DRAM.For the NAND business, despite sluggish demand in the first half, the company completed its development of 321-layer QLC products. SK hynix also achieved the highest annual revenue on record, responding to customer demand centered on eSSD in the second half.The company noted that as the AI market shifts from training to inference while demand for distributed architectures expands, the role of memory will become increasingly critical. Accordingly, not only demand for high-performance memory such as HBM is expected to grow continuously, but also for overall memory products including server DRAM and NAND as well.In response, SK hynix plans to further strengthen its proven quality, technological leadership and mass-production capabilities, based on the customer trust it has secured as the only industry player capable of stably supplying both HBM3E and HBM4 simultaneously. In particular, having successfully completed the preparation stages to mass produce HBM4 - for the first time in the industry - in September last year, the large-scale production of the next-generation HBM has been underway to meet customer requests, the company said.The company also sets to maintain its HBM4 leadership while strengthening customer and partner collaboration to supply optimized products in 'Custom HBM', which continues to gain traction as a key differentiator.For conventional DRAM, SK hynix intends to accelerate the transition to the 1cnm process, expanding its AI memory product portfolio to include solutions like SOCAMM2 and GDDR7. For NAND, the company plans to maximize product competitiveness by transitioning to 321-layer technology, while actively addressing AI data center storage demand by leveraging Solidigm's QLC eSSD.SK hynix stated that it will prioritize meeting customer demand amid supply-demand imbalances by reinforcing partnerships. To this end, the company plans to maximize production capacity of the M15X fab in Cheongju at an early stage. The company also intends to secure stable mid-to-long-term production capabilities through the construction of the first fab in the Yongin Semiconductor Cluster.The construction of the advanced packaging facilities in Cheongju and Indiana, the U.S., are also progressing smoothly. This enables the company to establish integrated global manufacturing capabilities spanning front-end and back-end processes, having able to respond flexibly to changes in customer demand.Meanwhile, backed by record-high financial performance, SK hynix announced a large-scale shareholder return program to enhance shareholder value.The company will deliver an additional dividend of 1 trillion won, equivalent to 1,500 won per share. Combined with the regular quarterly dividend of 375 won, the year-end dividend will total 1,875 won per share, bringing total dividends for FY2025 to 3,000 won per share, or 2.1 trillion won in aggregate.SK hynix also plans to cancel 15.3 million treasury shares (approximately 12.2 trillion won based on the closing price on the 27th) equivalent to 2.1% of total shares outstanding, signaling a long-term commitment to enhance per-share value and shareholder returns.Song Hyun Jong, President and Head of Corporate Center, said that SK hynix will continue to generate sustainable performance growth while maintaining the optimal balance between future investment, financial stability and shareholder returns, based on the company's technological edge. "We will strengthen our role not merely as a product supplier, but as a core infrastructure partner in the AI era, enabling customers to meet their AI performance requirements."4Q25 Financial Results (K-IFRS). Credit: SK hynixFY2025 Financial Results (K-IFRS).Credit: SK hynix

The Ministry of Economic Affairs (MOEA) officially announced the launch of the 2026 Best AI Awards during a press conference held on January 20, 2026. This prestigious competition serves as a strategic platform to bridge Taiwan's software innovation with its world-class semiconductor industry, aiming to solidify the nation's position as a global hub for artificial intelligence.Championing the Next Wave: Agentic AISpeaking at the launch, Director General Kuo Chao-Chung of the Department of Industrial Technology highlighted a pivotal shift in the industry. "AI is moving beyond traditional discriminative and generative models toward 'Agentic AI' - autonomous systems capable of independent planning, reasoning, decision-making, and task execution," Kuo stated. To encourage this evolution, the 2026 awards will prioritize projects focusing on Agentic AI and AI model lightweighting, ensuring that software solutions are both intelligent and hardware-efficient.Competition Highlights and Incentives The 2026 Best AI Awards are divided into two primary tracks: AI Applications and IC Design. Registration is open from now until March 16, 2026, with the final competition set for April 25, 2026.Financial rewards are substantial, with corporate winners eligible to secure up to NT$1 million. Student teams also have significant incentives, with potential prizes reaching up to NT$300,000.Beyond direct cash prizes, winning Taiwanese companies benefit from enhanced R&D support, receiving an additional 10% subsidy bonus when applying for the "A+ Industrial Innovative R&D Program." Furthermore, these awards serve as a springboard for startup growth by providing domestic winners with priority access to the Startup Terrace Kaohsiung.International teams and global companies are incentivized through specialized support designed to integrate them into Taiwan's technology ecosystem. In addition to competing for the top cash prizes, winning global entities receive streamlined eligibility for Taiwan’s Employment Gold Card, as well as priority access to Startup Terrace Kaohsiung to support residency applications. These benefits are designed to facilitate global talent recruitment while enabling international startups and professionals to establish operations, scale their businesses, and build a long-term presence in Taiwan.Finalists gain valuable market exposure by being showcased at major international platforms such as InnoVEX. This placement is intended to facilitate critical business matching and attract potential investment from global stakeholders in the AI and semiconductor industries.A Proven Track Record of SuccessThe inaugural edition of the awards attracted over 1,000 teams and has already yielded significant commercial results. Previous winners, such as Zhen-Sheng Semiconductor and ProsperX Medical Technology, have successfully raised NT$180 million and NT$100 million respectively following their participation. Established firms like Etron Technology also reported that the awards significantly boosted their visibility among global system manufacturers.The MOEA invites students, startups, and established enterprises to showcase their innovations and join Taiwan's thriving AI ecosystem. By fostering a "Software-Hardware Synergy," Taiwan aims to redefine the global semiconductor supply chain through AI-driven intelligence.For more information on registration and competition guidelines, please visit the official website or follow the official LinkedIn page for the latest updates.