As the AI computing race accelerates and semiconductor manufacturing capacity is increasingly redirected toward advanced memory, the memory industry is shifting from price-driven competition to a battle over resource allocation and supply security. The rapid expansion of high-bandwidth memory (HBM) production continues to consume advanced fabrication capacity, constraining the supply of conventional memory products and creating persistent shortages across the market. This concentration of high-end production has widened the supply-demand gap, driving the average selling prices of certain premium memory products to levels eight to ten times higher than previous norms

DATOTEK INTERNATIONAL CO., LTD., a Taiwan-based memory and storage manufacturer, showcased its latest AI-powered sustainable storage solutions and AI infrastructure technologies at COMPUTEX 2026 in Taipei under the exhibition theme "AI Together." The company presented a comprehensive portfolio of AI storage, edge AI computing, PCIe Gen5 SSD thermal management, USB4 portable SSDs, and next-generation memory products designed for both enterprise and consumer markets

Technology news, product launches and show highlights from the event attendees available for download (PDF).

As demand for AI servers, high performance computing (HPC), and data center infrastructure accelerates, the semiconductor and power electronics test equipment market is experiencing strong growth. This momentum is expected to continue as advanced semiconductor technologies drive increasingly sophisticated testing requirements.Chroma ATE Inc., a company with more than two decades of experience in automated test equipment (ATE), recently held its Semiconductor Test Equipment User Conference in Hsinchu, Taiwan. Registration quickly reached capacity, reflecting the company's growing prominence in the semiconductor testing market. In 2025, Chroma reported consolidated revenue of US$943 million and grew its global workforce to nearly 4,000 employees.Credit: ChromaAddressing the Testing Challenges of the AI Era with In-House InnovationGeorge Chang, President of Chroma's Semiconductor Test Equipment Business Unit (STE BU), emphasized that Chroma is one of the few ATE suppliers with the in-house capability to develop  application-specific integrated circuits (ASICs) for its own test systems. This capability creates a "technology moat" that reinforces Chroma's competitive advantage.To address the diverse needs of the semiconductor testing market, Chroma has built its portfolio around three specialized test platforms. The Chroma 3680 is designed for advanced devices used in AI, HPC, microcontroller unit (MCU), and mixed-signal applications, the Chroma 3650 addresses power devices and systems-on-chip (SoCs), and the Chroma 3380 targets mainstream MCUs, Internet of Things (IoT) devices, and other chips manufactured on mature process nodes. Combined shipments across the three platforms have exceeded 4,000 systems. This year, Chroma also introduced the compact Chroma 3650-S2C platform for high-power devices, responding to increasingly rapid product cycles across the semiconductor industry.Proprietary PINF ASIC Enables Seamless Migration for Complex SoC TestingAs AI and HPC drive more complicated chip designs, testing is becoming more diverse and technically more demanding. At the same time, the industry is facing growing pressure to reduce costs and shorter test development schedules. Spancer Lee, Senior Manager of product marketing at Chroma's STE BU, explained that the Chroma 3680 integrates Chroma's next-generation, self-developed PINF ASIC with a range of specialized system boards, each supporting different test functions. This architecture enables comprehensive coverage across four major testing domains: digital, analog, mixed-signal, and RF devices. Chroma plans to continue expanding the platform's capabilities, leveraging the flexibility of its in-house hardware development to support high-precision testing of increasingly complex and high-frequency SoC devices. To accommodate the intensive data transfer and processing requirements involved with testing advanced semiconductor devices, the Chroma 3680 employs an 8-lane, 20 Gbps parallel high-speed data architecture. Optimized firmware enables high-speed transfer and more efficient processing of complex test data. The architecture also minimizes synchronization latency between digital and analog resources, improving resource integration for highly complex SoC test programs.The Chroma 3680 architecture is also compatible with certain probe card models used on competitor platforms. This allows customers to retain existing load boards and test programs without modifying their hardware, reducing migration costs while maximizing ROI and helping chipmakers gain a competitive edge in the market.Credit: ChromaChroma 3650-S2C Delivers High-Voltage Capability in a Compact FootprintWith NVIDIA signaling a transition toward direct 800V DC power delivery for server racks, semiconductor manufacturers face growing demand for high-voltage testing. To address this need, Chroma developed the 3650-S2C, a new test platform optimized for high-power semiconductor devices including silicon carbide (SiC) and gallium nitride (GaN). Through extensive engineering refinement, the platform's slot count was reduced to six, optimized for the requirements of intermediate-level testing. Chroma also substantially streamlined the external high-voltage power rack by integrating it directly into the test head. The 3650-S2C cuts the system footprint by half, enabling seamless integration with mainstream turret handlers for volume production while helping OSAT partners save valuable factory space and infrastructure costs.Aaron Chuang, Senior Manager of Advanced Technology Development at Chroma's STE BU, highlighted two key characteristics of the Chroma 3650-S2C: powerful high-voltage capability and a compact footprint. Building on the proven Chroma 3650-S2 platform, the solution integrates a high-voltage source-measurement unit (HVSMU) board to deliver a comprehensive, one-stop solution for five critical test parameters. It enables customers to safely and accurately overcome the 4 kV test threshold while maintaining production throughput, providing a powerful foundation for next-generation high-voltage power semiconductors to advance into mass production.Credit: ChromaIntegrated High-Density RF Technology Addresses Next-Generation Ultra-High-Frequency Wireless TestingThe rapid adoption of AI, edge computing, and intelligent connected devices continues to drive growth across Wi-Fi, Bluetooth, GNSS, IoT, mobile, and wearable applications. At the same time, wireless communication chipsets are undergoing a significant generational upgrade.To address these evolving requirements, Chroma has integrated its new HDRF3 (high-density radio frequency) test module into the Chroma 3680 platform. The solution is designed to support the increasing bandwidth, advanced modulation schemes, and multi-antenna architectures required by next-generation wireless technologies. This positions the company to capture the high-performance networking and AIoT chip packaging and testing markets.According to Jackson Hsu, associate director of application engineering at ADIVIC, a Chroma-affiliated company, Wi-Fi 7 expands channel bandwidth from 160 MHz to 320 MHz while increasing modulation density to 4096-QAM. These advances place exacting demands on RF performance and noise characteristics in test instrumentation. The HDRF3 test module was specifically developed to meet these requirements, extending operating frequencies up to 8 GHz and providing full coverage for Wi-Fi 7 as well as future Wi-Fi 8 implementations.ADIVIC also plans to launch an extension model, the 35809, featuring a simplified 16-port design while maintaining complete software compatibility with the production-oriented HDRF3 platform. The solution is targeted at applications requiring RF IC testing only. Chroma's in-house ASIC expertise allows the company to adapt the hardware architecture of its Chroma 3680, 3650, and 3380 platforms to distinct market requirements. From the space-efficient Chroma 3650-S2C to the high-density HDRF3 module, each development addresses emerging technical challenges across the semiconductor industry.Credit: ChromaThrough a combination of flexibility, high return on investment, and seamless platform migration strategies, Chroma helps customers accelerate adoption while reducing operational barriers. As AI, advanced computing, wireless connectivity, and power semiconductors continue to reshape the industry, Chroma is positioning itself as a long-term strategic partner for the world's leading semiconductor manufacturers.

SK hynix Inc. (or "the company", www.skhynix.com) announced today that its Board of Directors has approved a resolution to repurchase and fully cancel 40 trillion won worth of its own shares. The Company also revealed plans to return "over 50% of cumulative Free Cash Flow (FCF)" generated during the program period (2025-2027), adopting a framework centered on the parallel execution of share repurchases and cancellations alongside cash dividends.Following the Board meeting held earlier today, the Company formally disclosed the share repurchase and cancellation resolution along with the shareholder return plan via a regulatory filing.The decision stems from the assessment that the Company's intrinsic value—underpinned by its business competitiveness, robust cash generation capability, and mid-to-long-term growth potential—is not fully reflected in its current stock price.SK hynix continues to record record-breaking financial performance, maintaining its leadership in the AI memory market. As of the end of the second quarter of this year, the Company's net cash stood at approximately 69 trillion won, demonstrating significantly improved cash generation capability.The total planned amount for the share repurchase is 40 trillion won. Based on the closing stock price of 1,662,000 won on the day prior to the Board resolution, this represents approximately 24.07 million shares, or about 3.3% of total issued shares (730,492,365 shares). The repurchase period is scheduled to run for approximately three months starting August 20, with all repurchased shares set to be cancelled upon completion of the acquisition.This initiative accelerates the execution of the existing shareholder return program ahead of schedule. In November 2024, SK hynix announced a program to execute shareholder returns within the range of 50% of its three-year (2025-2027) cumulative FCF, noting that it would consider early execution prior to the program's expiration if FCF increased meaningfully due to improved operational performance.The 40 trillion won program marks the largest treasury share cancellation ever conducted by a South Korean listed company. The Company noted that progress toward its financial health targets remains on track, affirming its commitment to maintaining a stable financial structure while delivering sustained shareholder value.Through today's regulatory filing, SK hynix announced plans to pursue an expansion of its total shareholder return target from the previous "within 50% of cumulative FCF" to "over 50% of cumulative FCF". Under the framework, shareholder returns will be executed through a dual-track approach running share repurchases/cancellations alongside cash dividends, while options to expand payouts—including the existing fixed dividends and special dividends—are also under consideration.SK hynix emphasized that it intends to pursue additional shareholder returns within the program period through the parallel execution of share repurchases/cancellations and dividends, by taking into account cash flows, market conditions, and distributable profits, adding that specific details regarding scale and execution will be announced following Board approval at the time of its third-quarter earnings release.

The AI industry is entering a new phase centered on inference efficiency and agentic deployment. As AI search, enterprise copilots, autonomous driving, embodied intelligence, AI PCs, AI smartphones, and multi-agent collaboration accelerate, the token has evolved beyond a simple unit for measuring model calls to become a core metric for AI infrastructure efficiency, cost structure, and production capacity. Amid this shift, the value of memory and storage is being redefined.From model loading and KV cache to RAG, vector retrieval, long-context inference, and data accumulation, every critical stage of AI inference and application depends on memory and storage that are high-performance, highly reliable, low-latency, and built to scale sustainably. AI memory and storage are evolving from back-end data repositories into core foundations underpinning token generation, caching, retrieval, scheduling, and data accumulation.At the same time, AI workloads are extending from cloud data centers to edge nodes and end devices, giving rise to a new paradigm of coordinated cloud-edge-device evolution. Breakthroughs in AI-era memory and storage will no longer come from single-point performance gains alone, but from system-level collaboration, ecosystem integration, and supply chain resilience; industry collaboration is moving beyond linear upstream-downstream supply relationships toward a new paradigm of deep alignment and shared growth, built around cloud-edge-device workload demands, joint product definition, scenario validation, supply security, and shared value creation.Against this backdrop, the 5th GMIF Innovation Summit (Global Memory Innovation Forum 2026), themed "The Future Built on AI Memory and Storage," will be held on September 22–23, 2026, in Shenzhen, China. Rooted in the memory and storage supply chain and geared toward new AI infrastructure and cloud-edge-device scenarios, the summit will bring together leading voices across the ecosystem, including players in wafer manufacturing (IDMs), controllers, modules, OSAT, equipment and materials, servers and systems, AI computing, AIDC, cloud services, smart terminals, automotive electronics, and embodied intelligence, along with investment and academia institutions, to explore the trends, demands, and collaborations shaping memory and storage in the AI era.Summit Topics: Exploring New Variables in the Token Economy through the Lens of AI Memory and StorageCentered on the memory and storage industry chain and framed through the lens of the token economy, GMIF 2026 will build a cross-sector exchange platform around AI infrastructure efficiency, cloud-edge-device application expansion, industry ecosystem collaboration, and capital value reassessment.Token Economy: Redefining the Value of Memory and Storage for AI. Examining how memory and storage are evolving beyond capacity resources into critical infrastructure for Token-driven AI, with a focus on AI inference, KV cache, vector retrieval, RAG, long-context inference, and data persistence.Token Economy: Redefining the Value of Memory and Storage for AI. Examining how memory and storage are evolving beyond capacity resources into critical infrastructure for Token-driven AI, with a focus on AI inference, KV cache, vector retrieval, RAG, long-context inference, and data persistence.Cloud-Edge-Device Collaboration: Extending AI Memory and Storage Across New Application Frontiers. Exploring the upgrade of memory architectures and product innovation opportunities driven by diverse AI-demanded scenarios, including AI data centers, edge inference, AI PCs, AI smartphones, AI glasses, intelligent vehicles, embodied intelligence, and multi-agent systems.Global Landscape: Building an Open, Resilient Ecosystem Together. Addressing shifting global industry dynamics and supply chain restructuring, with a focus on data governance, supply diversification, regional collaboration, and long-term sustainability.Industry Capital: Unlocking the Long-Term Value of AI Memory and Storage. Bringing together investment institutions, industry funds, securities firms, listed companies, and innovative enterprises to jointly assess industry cycles, capital logic, and long-term value.Centered on the Memory and Storage Industry Chain, Connecting the New AI Infrastructure EcosystemAs a leading global platform for the memory and storage industry, GMIF connects wafer IDMs, controllers, modules, OSAT, equipment and materials, terminal companies, and channel partners to drive technical exchange and ecosystem collaboration.GMIF connects IDMs, OSAT, controllers, cloud, equipment, & terminals for memory. Credit: SMIAFor 2026, GMIF expands further into compute providers, cloud and AIDC services, large model platforms, the RAG/agent-layer, smart terminals, automotive electronics, and embodied intelligence, strengthening ties between memory & storage and AI infrastructure across cloud, edge, and device.Summit Programs: Keynotes, Rankings, and a Closed-Door Investment ForumThe summit combines keynotes, roundtables, supply-demand matchmaking, product showcases, industry rankings, awards, and global livestreaming.GMIF Innovation Summit (main forum): keynotes from leaders in memory and storage, computing, cloud services, terminals, academia, and investment, focused on AI memory and storage technology, market opportunities, cloud-edge-device expansion, and token efficiency.AI Industry and Investment Forum: a closed-door, invitation-only session on AI memory and storage, enterprise SSDs, compute infrastructure, edge and on-device AI, advanced packaging, and industry M&A.Summit features keynotes, matches, showcases, awards, rankings, & livestreams. Credit: SMIAGMIF 2026 will also present its annual awards across Technology Innovation, Market Services, Industry Contribution, AI Infrastructure, Terminal Applications, and Capital Value. Alongside the awards, the Shenzhen Memory Industry Association — together with the School of Integrated Circuits at Peking University, JWinsights, and other institutions — will release the Memory & Storage Industry Rankings 2026, spanning semiconductor memory & storage, AI optical interconnect, semiconductor equipment, and AI computing.GMIF 2026 presents annual awards and industry rankings for storage & AI. Credit: SMIAA Track Record of Impact: Proven Value as an Industry PlatformSince its launch in 2022, GMIF has grown into a leading annual platform for the global memory & storage industry, connecting wafer IDMs, controller and module makers, OSAT providers, equipment and materials suppliers, terminal companies, investors, and media. GMIF2025 brought together 42+ key companies, showcased 180+ innovative products, presented 37 annual awards, and drew over 1,600 on-site attendees — with total exposure exceeding 420 million impressions and international livestream viewership surpassing 700,000.In 2026, GMIF builds on that momentum, expanding its reach across AI infrastructure, cloud-edge-device applications, and industry capital as it evolves from an annual exchange into a full AI memory & storage industry collaboration platform.Full-Spectrum Media Reach, Amplifying the Industry's VoiceGMIF 2026 will run an integrated communications campaign — features, short video, online and offline advertising, and outreach through 50+ central, financial, and semiconductor/memory vertical media outlets — to bring the AI memory & storage conversation to decision-makers, technologists, and investors alike.Full-Spectrum Media Reach, Amplifying the Industry's Voice. Credit: SMIALet AI memory and storage be our starting point — together, let's explore the new value the token economy brings to the industry, and witness memory and storage step onto the center stage of AI infrastructure.Join us September 22–23, 2026, as the 5th GMIF Innovation Summit convenes in Shenzhen.Registration is now open. We welcome industry professionals, technology leaders, and ecosystem partners to join us. Registration, awards submission and for more information about GMIF, please visit.