Rebellions Inc., a leading AI semiconductor company based in South Korea, today announced a collaboration with Marvell Technology, Inc. (NASDAQ: MRVL) to offer high-performance, energy-efficient AI systems tailored for regionally driven and sovereign-backed AI initiatives across APAC and the Middle East.As AI infrastructure becomes increasingly central to national competitiveness, a shift is underway from standardized, GPU-based architectures toward domain-specific systems built around custom ASICs. This change is being driven not only by global hyperscalers but also by sovereign-backed initiatives and regional cloud providers seeking infrastructure that delivers scalability, efficiency, and control.This collaboration will enable Rebellions to design customer-specific AI accelerators utilizing Marvell® custom platforms—leveraging advanced packaging, high-speed SerDes, and die-to-die interconnects, resulting in a tightly integrated, end-to-end rack solution built for high-performance, energy-efficient AI inference at scale."AI infrastructure is undergoing a fundamental shift—organizations are no longer settling for one-size-fits-all solutions," said Sunghyun Park, CEO of Rebellions. "By collaborating with Marvell, we're combining deep expertise in AI chip design with cutting-edge silicon integration to deliver custom AI infrastructure designed to meet the real-world needs of sovereign entities.""Custom AI infrastructure is key to unlocking the next wave of data center innovation," said Will Chu, Senior Vice President and General Manager of Custom Cloud Solutions at Marvell. "We look forward to collaborating with Rebellions to offer next-generation custom AI infrastructure tailored for performance, efficiency, and scalability."About RebellionsRebellions develops AI accelerators optimized for Large Language Models and large-scale inference, delivering industry-leading energy efficiency. Its flagship chip, REBEL, uses chiplet architecture and 144GB HBM3E memory for massive-scale AI inference. The platform includes software for seamless data center integration. Building on its AI accelerator ATOM's proven mass-production experience since 2024, Rebellions is now backed by strategic investors including SK Telecom, SK hynix, Aramco's Wa'ed Ventures, and KT. Strengthened by its merger with SK SAPEON, Rebellions is Asia's leading independent AI semiconductor platform.
IEC 62443 defines procedures for implementing electronic system security in Industrial Automation and Control Systems (IACS). In response to the growing trend of smart industrial control systems, both industry and governments have introduced cybersecurity regulations and standards. One of the most frequently referenced standards is IEC 62443, proposed by the International Electrotechnical Commission (IEC).Originally developed by the International Society for Automation (ISA) and published by the American National Standards Institute (ANSI), the standard was later adopted by the IEC. Initially known as ANSI/ISA-99 or ISA99, it is now published as ISA-62443-x-y by ISA and IEC 62443-x-y by IEC.Cybersecurity issues in industrial control systems are closely related to the level of computerization or intelligence of the system. Recent government reports indicate that hackers are increasingly targeting critical infrastructure and high-tech manufacturing sectors. They infiltrate external networks and gradually penetrate the networks housing industrial control systems, creating a ripple effect. Preventing cybersecurity vulnerabilities in upgraded systems is a major challenge before transitioning to Industry 4.0. These issues can affect product quality, public health, and safety, potentially causing significant damage.IEC 62443 addresses these concerns by providing security standards for IACS to tackle a wide range of cybersecurity challenges.Overview of IEC 62443The technical specifications of this standard apply to all users of IACS, including organizations involved in operations, maintenance, engineering, component manufacturing, and those affected by or involved in control systems. It emphasizes the importance of collaboration between IT and operations and the role of engineering and manufacturing in ensuring enterprise-wide security. The standard is also crucial for network administrators working with IACS. Here comes a structured figure of the standard for reference and simplified elaboration as follows.Figure 1: IEC 62443 Standard Level CategorySource: IECIEC 62443 is structured into four main categories:1. General: Covers foundational concepts, terminology, and methodologies. It explains the threats and issues in industrial automation systems and introduces a reference model to help organizations analyze and implement protective mechanisms.2. Policies & Procedures: Describes the IT security management system and requirements for asset owners (e.g., factory managers), outlining how to ensure operational safety through management policies and processes.3. System: Provides technical specifications from the perspective of system integrators. It discusses how to ensure integrated automation solutions are secure from cyber threats, including conducting security assessments.4. Component: Focuses on the design and development requirements for control system components. It guides product suppliers on how to develop secure products and ensure the safety of automated equipment.Goals of IEC 62443 and Overview of IEC 62443-4 Certification Requirements and Device CategoriesTable 1: IEC 62443 standard for product design flow and certification requirementsSource: IECIEC 62443 certification involves evaluating, integrating, and maintaining specific product functions or solutions under development. The IEC 62443-4 standards apply to asset owners, system integrators, product suppliers, and certification bodies (including government and regulatory agencies).- System Integrators use the standard to procure IACS-compliant control system components and define the required security levels (SL-T) for different zones.- Product Suppliers use the standard to understand the security level requirements (SL-C) for their components. Even if a component doesn't provide all required capabilities, it may still meet requirements when integrated with higher-level systems.IEC 62443-4 defines component requirements (CR), derived from system requirements (SR) in IEC 62443-3-3, which in turn are based on foundational requirements (FR) from IEC 62443-1-1. CRs may include requirement enhancements (RE), and the combination of CR and RE determines the achievable security level.The four types of component requirements are:• Software Application Requirements (SAR)• Embedded Device Requirements (EDR)• Host Device Requirements (HDR)• Network Device Requirements (NDR)Winbond's Secure Flash Memory SolutionsWinbond, a global leader in semiconductor memory solutions, offers secure flash memory products that meet stringent cybersecurity certification requirements. These high-performance, secure flash memories support:• Secure Boot• Secure Storage• Post-Quantum Cryptography (PQC) authenticated firmware updates• Platform Integrity VerificationThey comply with global security standards such as NIST 800-193, NIST 800-208, EN18031, and are certified under Common Criteria and SESIP (Security Evaluation Standard for IoT Platforms).Winbond's secure flash memory products, W75F, W77Q, and W77T, support the full range of IEC 62443 Security Levels (Target), aligning with major cybersecurity standards.Figure 2: Cybersecurity StandardsSource: WinbondConclusionWinbond's secure flash memory solutions help system manufacturers meet regulatory compliance, enhance platform security, and defend against hardware-based cyberattacks. For more information, visit Winbond Secure Flash Memory or download the latest Hardware Security White Paper..
SK Hynix Inc. (or "the company") announced today that it has recorded KRW22.232 trillion in revenues, KRW9.2129 trillion in operating profit(with an operating margin of 41%), and KRW6.9962 trillion in net profit(with a net margin of 31%) in the second quarter.Both revenues and operating profits stood at all-time highs, beating the previous best results in the fourth quarter of last year.The company (www.skHynix.com) said that an aggressive investment by global big tech companies into AI led to a steady increase in demand for AI memory. Shipments of both DRAM and NAND flash were higher than expected, helping the company log the best quarterly results.SK Hynix expanded sales of 12-high HBM3E in the DRAM space, while registering a growth in sales of NAND for all applications. With the industry-leading competitiveness in AI memory and profitability-first management discipline, the company has maintained a positive earnings trend.With the latest financial results, the company's cash and cash equivalents increased to KRW17 trillion at the end of June, up by KRW2.7 trillion from a quarter earlier. Its debt ratio and net debt ratio stood at 25% and 6%, respectively, as net debt fell by KRW4.1 trillion, compared with the previous quarter.In the second quarter, the inventory level was kept steady as customers increased both memory orders and production of their finished products. Demand for memory is expected to continue to grow as customers plan to launch new products in the second half.Particularly, SK Hynix foresees that increasing competition among big tech companies to enhance inference of AI models would lead to higher demand for high-performance and high-capacity memory products. Ongoing investments by nations to build sovereign AI* would also help generate long-term demand for memory, it said.* Sovereign AI: a nation's ability to build an independent AI system with its own infrastructure, data, talent, and business networkThe company expects the solid performance of its products and mass-production capabilities to help double HBM, compared with a year earlier, to generate stable earnings. It will also ensure the timely provision of HBM4 in accordance with customers' requests to remain competitive.SK Hynix will start the provision of an LPDDR-based module for servers within this year, and prepare for GDDR7 products for AI GPUs with an expanded capacity of 24Gb from 16Gb in a bid to enhance its leadership in the AI memory market with product diversification.For the NAND business, the company will maintain a prudent stance for investments, considering demand conditions and profitability-first discipline, while continuing with product developments in preparation for improvements in market conditions. Particularly, it will expand sales of QLC-based high-capacity eSSD and build a 321-high NAND-based product portfolio to enhance its market leadership.Song Hyun Jong, President and Head of Corporate Center, said that SK Hynix will carry out part of the planned investments preemptively this year for the smooth provision of major products with visible demand for next year, including HBM. "We are on track to meet our goal as a Full Stack AI Memory Provider, satisfying customers and leading market expansion through the timely launch of products with best-in-class quality and performance required by the AI ecosystem."2Q25 Financial Results (K-IFRS), Unit: Billion KRW 2Q251Q25Q/Q2Q24Y/YRevenues22,23217,639.126%16,423.335%Operating Profit9,212.97,440.524%5,468.568%Operating Margin41%42%-1pp33%8ppNet Income6,996.28,108.2-14%4,12070%*Note: Financial information on earnings is based on K-IFRS*Note: Please note that the financial results discussed herein are preliminary and speak only as of July 24, 2025. Readers should not assume that this information remains operative at a later time.
The development of quantum computing is fundamentally reshaping the rules of cybersecurity. Traditional public key encryption mechanisms, such as RSA and ECC (Elliptic Curve Cryptography), rely on the complexity of problems like prime factorization and discrete logarithms for their security. However, what is extremely difficult for classical computers to break security could be effortlessly solved by quantum computers.Background and Challenges of Post-Quantum Cryptography (PQC)This is no longer a distant threat. Tech giants such as Google, IBM, and Intel have already demonstrated prototype quantum processors and continue to expand their qubit performance and maintain system stability. While today's low-qubit quantum computers still cannot break commonly used encryption algorithms in real-time, the cybersecurity community has already entered a stage of "quantum defense" deployment.For government agencies and critical infrastructure, this is an urgent concern. The US National Institute of Standards and Technology (NIST) has launched a Post-Quantum Cryptography (PQC) standardization project and announced the first batch of selected encryption algorithms in 2022. This signals that enterprises and device manufacturers must begin considering long-term protection strategies for firmware, data, and keys under quantum threats.Challenges in the PQC Transition: Firmware and Storage Devices at the ForefrontIn many systems, sensitive data and cryptographic keys are often stored in external non-volatile memory, such as SPI NOR Flash. For example, in embedded devices, servers, and network equipment, these components are responsible for storing firmware and critical decryption keys. Once compromised, the entire platform's trust foundation is at risk.However, most external memory on the market today only offers basic write protection and lacks features such as active encryption, region-specific authorization, or key management capabilities - let alone the computational support and security architecture required for increasingly complex PQC algorithms.In addition, many devices face the following challenges:*Difficulty integrating PQC validation and secure storage mechanisms into existing architectures*Lack of storage components that support multi-stage firmware validation and regional isolation*Firmware storage areas are susceptible to forgery or rollback attacks by adversaries.Therefore, integrating "Secure Flash that meets PQC-era requirements" into existing systems has become a critical step in advancing cybersecurity.Winbond TrustME Secure Flash - W77Q: A PQC-Ready Security SolutionIn response to trends in post-quantum and firmware security, Winbond has developed the TrustME Secure Flash - W77Q series. Designed specifically for Platform Firmware Resiliency (PFR) and hardware-based root of trust, the W77Q also supports the hash-based LMS (Leighton?VMicali Signature) post-quantum signature algorithm, providing a complete and robust security solution for device manufacturers.Highlight 1: Multi-Level Access and Key ManagementW77Q implements hardware-based region-level access control and key authentication systems. It enables the creation of multiple independent memory partitions, each configured with its own dedicated key and encryption rules. This is especially important when integrating PQC encryption workflows, and supports:*Independent hardware verification of each firmware stage (Pre-boot, BMC, UEFI, etc.)*Protection against rollback attacks*Restriction of access to specific memory regions (enforced for authorized users only)This architecture not only protects existing firmware structures but also provides a secure and scalable framework for integrating future PQC key management.Highlight 2: Mirror Design with Auto-Recovery Support (PFR-ready)W77Q features a built-in Recovery image region and supports automatic switching and verification. If the primary firmware becomes corrupted or fails validation, the system can automatically switch to the Golden Image without external intervention, achieving Platform Firmware Resiliency (PFR).This design aligns with the established Protect?VDetect?VRecover architecture is defined by NIST SP 800-193 and integrates seamlessly with major platform root of trust implementations.Highlight 3: LMS Key Pre-Provisioning and HSM Injection to Strengthen Supply Chain Root of TrustIn addition to its built-in secure zones and key management capabilities, W77Q supports the pre-injection of LMS post-quantum key pairs. Winbond can assist OEMs in securely embedding LMS public-private key pairs during wafer test or pre-packaging stages, ensuring each device has a unique and traceable cryptographic identity.The injection of the LMS private key is particularly important: it allows for secure device attestation and traceability. By embedding a unique private key within each W77Q device, only the original device can generate valid LMS signatures. This enables downstream systems to verify both the authenticity of the device and the integrity of its firmware, ensuring that only authorized firmware is executed and that the device is genuine. Such traceability and attestation are essential for establishing a robust root of trust throughout the supply chain, significantly enhancing cybersecurity and mitigating risks of device cloning or unauthorized firmware modification.Furthermore, Winbond supports key injection using Hardware Security Modules (HSMs), offering stronger key protection and verifiable injection processes. This functionality enables customers to establish a secure root of trust during chip manufacturing and assembly, further enhancing supply chain security.A One-Stop Secure Storage Solution to Accelerate the Adoption of Future Security ArchitecturesThe true value of W77Q lies not only in solving the "secure memory" problem, but also in addressing many of the supporting tasks developers would otherwise have to manage themselves:*Ready-to-use APIs for easy integration with firmware verification workflows*Secure key storage and region-level access control are handled by W77Q hardware, saving host resources*Support from a full-fledged security team to provide ISO 26262 and NIST PFR compliance documentation and test reports*Pin-to-pin compatible with standard SPI Flash - no need for additional hardware space, enabling low-cost integrationBy supporting both ISO 26262 for functional safety and ISO/SAE 21434 for automotive cybersecurity, W77Q helps manufacturers address both safety and security compliance requirements in modern embedded and automotive systems.For teams looking to implement functional safety, PQC, or platform firmware resiliency, this is an ideal "security accelerator."Conclusion: Secure Flash Is the First Line of Defense in the Post-Quantum Security EraWith the rollout of NIST PQC standards and the ongoing progress of quantum computing, device manufacturers and system integrators can no longer ignore the risks at the firmware level. Winbond's TrustME Secure Flash - W77Q series offers more than just memory - it provides a foundation of trust that supports system evolution, compliance, and long-term security maintenance.Whether you're an embedded platform developer or a module designer in automotive or industrial control sectors, adopting next-generation Secure Flash aligned with PQC and functional safety standards will be a key step in your cybersecurity transformation journey.To learn more about Winbond's advanced security solutions, visit Winbond's website, contact Winbond directly, or download the latest Hardware Security White Paper.
The Transport Department has given vehicle owners across Maharashtra one final window to upgrade their old number plates. Any car or bike registered before 1 April 2019 must display a High Security Registration Plate (HSRP) before 15 August 2025. From 16 August onward, authority teams may start the inspection across the state, and officials have made it clear that no further extensions will follow. The directive applies to roughly 2.10 crore vehicles, yet only about 23 lakh have switched so far. That gap is large enough for authorities to issue a firm warning that no further extensions will be issued. Read on to learn more about the compliance and the process.Why does HSRP matter?HSRP is more than just a new number plate. It carries built-in security elements that help curb fraud and theft.*Chromium hologram: A small Ashoka Chakra symbol in the top-left corner shatters if anyone tries to peel the plate away.*Laser-etched code: A unique string of numbers links the plate to the central vehicle register, making cloning extremely difficult.*Snap locks: Special rivets break on the first attempt at removal, so a swapped plate cannot be reused without visible damage.Extension timelineThe Transport Department has already relaxed the schedule three times. The order originally announced 31 March 2025 as the cut-off, then shifted to the end of April, later to the end of June, and now to 15 August 2025. Officials say this is the last extension for the HSRP Maharashtra. A confirmed booking made before the deadline counts as proof of intent even if the fitting slot falls afterwards, shielding you from penalties.Risks of Missing the 15 August Cut-offThe following may be the key consequences of negligence:*Fines or seizures under the Motor Vehicles Act, once inspections begin.*Insurance complications if the vehicle is stolen or damaged; some claims may be delayed or rejected.*Lower resale value because buyers may increasingly expect a legally compliant plate.Three-step HSRP booking through ACKOHere are the key steps you will need to follow while booking your HSRP number plate online through ACKO:*Enter your registration number: Visit ACKO's HSRP page and type in the vehicle number. The portal automatically directs you to the authorised supplier for your city or district.*Add key details: On the supplier's site, fill in the engine and chassis numbers from the Registration Certificate. Choose either a centre visit or, if offered, a doorstep installation slot.*Pay and confirm: Complete the online payment, pick a time slot, and keep the confirmation message. If your chosen slot falls after 15 August, that digital receipt may protect you from fines.Tips for a smooth installation of HSRPHere are key tips for a smooth process:*Book early: Appointments fill quickly as deadlines approach.*Carry essential documents. Bring the Registration Certificate and a photo ID on installation day.*Check before you leave: Make sure the hologram is intact, the laser code matches your papers, and the plate sits straight.*Store the receipt: Keep a digital copy on your phone and a printout in the glove box until the transport database updates.Why is getting an HSRP plate immediately important?Here are the key reasons why you should act now:*Avoid penalties: Timely compliance prevents fines, seizures, and needless hassle.*Deter theft: The secure plate lowers the risk of number cloning and illegal resale.*Better slot choice: Early birds may be able to pick convenient time slots and skip long queues.Final wordChanging to an HSRP is a one-time task that usually takes less than an hour, but the benefits last for the life of the vehicle. Maharashtra's Transport Department has drawn a clear line: 15 August 2025 is the final date, and inspections start the very next morning. Book through ACKO today, lock in your HSRP plate and drive with peace of mind when enforcement begins. Waiting only invites crowded centres, higher stress, and the real risk of penalties. The most straightforward path is open right now; use it before the window closes.
Vietnam is changing fast. The country is attracting more foreign investment, especially in manufacturing and artificial intelligence (AI). Big companies from the US, Europe, and Asia are setting up factories and research centers here. This is not just about cheap labor anymore. It's about technology, skills, and new ideas.Foreign investment keeps risingIn the first five months of 2025, Vietnam pulled in over US$18 billion in foreign investment. That's a big jump from last year. Most of this money is going into factories, digital technology, and AI. Real estate is still important, but the focus is shifting. Investors want to build things, not just buy land.The government is making it easier for companies to set up shop. They're fixing rules, building better roads, and supporting new businesses. They want Vietnam to be a place where high-tech companies can grow. This is part of a bigger plan to boost the economy and create more jobs.Why companies are choosing VietnamVietnam's location helps. It's close to China, Japan, and other big markets. Labor costs are lower than in China. The country is part of several trade deals, so it's easier to export goods. These things make Vietnam attractive for companies that want to build electronics, semiconductors, and other tech products.Big names like Samsung, Intel, and Foxconn are already here. Apple is moving some of its manufacturing from China to Vietnam. Google and Meta are expanding too. Even SpaceX and the Trump Organization have announced big investments.But it's not just about foreign companies. Local firms like FPT are growing fast. They're working with global tech leaders like NVIDIA to build AI factories and train workers.AI is more than a buzzwordAI is not just a trend in Vietnam. It's becoming part of daily life and business. The government has set up a national AI strategy. They want to train more people, build research centers, and support startups. The goal is to make Vietnam a leader in AI in Southeast Asia.Vietnam now ranks 5th in ASEAN for AI readiness and 59th in the world. That's a big jump in just a few years. The country is investing in data infrastructure and encouraging universities to teach AI skills. FPT, for example, is working with NVIDIA to build an AI factory. They want 3-5% of the population working in AI in the future.Real-world AI applicationsAI is already making a difference in several areas:*Manufacturing: Factories use AI to manage supply chains, reduce waste, and improve quality. This helps them compete globally. *Healthcare: AI tools help doctors diagnose faster and more accurately. Startups are building AI systems for medical imaging and patient care. *Agriculture: Farmers use AI to monitor crops, predict weather, and manage resources. This boosts yields and cuts costs. *Education: Schools are using AI to personalize learning. Universities are launching new AI courses and research programs. *E-commerce: Online shopping is booming. AI helps companies manage logistics, recommend products, and serve customers better.The semiconductor pushSemiconductors are a big deal. Vietnam wants to be a hub for chip manufacturing, and the industry is expected to be worth over $20 billion by the end of 2025. Companies from the US, Japan, and other countries are building factories here, for example, Amkor just opened a $1.6 billion plant in Bac Ninh. The US and Vietnam have signed agreements to develop the chip industry together.Training workers is a priority. Vietnam aims to train 50,000 semiconductor professionals by 2030. FPT corporation, one of Vietnam's largest technology companies, plans to train 10,000 of them. They have set up new university programs and are working with partners from India and the UK.E-commerce and digital growthVietnam's e-commerce market is now one of the top three in Southeast Asia. In 2024, online sales are expected to hit $16 billion. More people are shopping online, using smartphones, and paying digitally. Big platforms dominate, but smaller websites and social media shops are growing too.AI is helping e-commerce companies run smoother. It's used for everything from managing warehouses to chatting with customers. Experts think AI could add up to $130 billion to Vietnam's economy by 2040, mostly through better productivity and new products.Challenges to watchVietnam's growth story is impressive, but there are real challenges.*Infrastructure: Some rural areas still lack good internet and tech access. This makes it hard for everyone to benefit from new technology. *Skills Gap: There aren't enough trained workers in AI and advanced manufacturing. More training and education are needed. *Regulation: The government is working on better rules for investment and data privacy. But it takes time to get it right. *Origin Fraud: Some companies use Vietnam as a stopover to avoid tariffs, especially from China. This can cause trade problems. The government is trying to crack down on this. *Environmental Concerns: Not all investments are good. Projects that pollute or don't meet safety standards are being watched more closely.What's next for Vietnam?Vietnam is not just a place for cheap assembly anymore. It's becoming a center for high-tech manufacturing and AI. The government, local companies, and foreign investors are all pushing in the same direction. They want to build a strong, modern economy that can compete globally.The country is setting up an international financial center to attract more investment. They are also working to make the business environment more transparent and reliable. This should help build trust and bring in more long-term partners.What Vietnam's growth means for youVietnam's rise in manufacturing and AI is real. The country is attracting more investment, building new factories, and training more people. AI is moving from buzzword to reality, with real impact in factories, hospitals, farms, and schools. The semiconductor industry is growing fast, and e-commerce is booming.But there are still hurdles. Infrastructure, skills, and regulation need more work. The government and businesses know this and are taking steps to fix it.Fusion Worldwide has an office right here in Vietnam. That means we are on the ground, ready to help you source what you need, navigate local markets, and make decisions that support your long-term goals. When you work with us, you are gaining a partner who understands the landscape, speaks the language and is invested in your success.The opportunity isn't just to react to change, but to shape your future supply chain and technology strategy. With the right support, you can turn Vietnam's momentum into your own advantage, today and for the long run.Tony Leong, Fusion Worldwide Senior Director of Business Development, Asia
With the ongoing evolution of automotive electrification and intelligence, functional safety has expanded from advanced ADAS platforms to every electronic control unit (ECU) and module component in the vehicle. Traditionally, these lower-level modules (such as lamp controllers, HVAC modules, seat adjustment modules, etc.) have not emphasized functional safety design due to cost sensitivity and relatively lower application risks.However, as vehicle manufacturers (OEMs) increasingly focus on the integration risks of entire vehicle systems, and Tier 1 suppliers gradually include ISO 26262 compliance as a key evaluation criterion during bidding, ECU and module products are facing unprecedented pressure to upgrade. Many products that were not originally designed with functional safety in mind are now required to provide evidence of safety design in order to be integrated into the overall vehicle safety architecture.For small and medium-sized module suppliers, this is both a challenge and an opportunity. Those who can quickly implement functional safety without significantly increasing development costs and timelines will be the first to gain market recognition.ISO 26262 is an international standard for the functional safety of electrical and electronic systems in road vehicles. Its primary goal is to ensure that automotive electronic and software systems do not lead to unacceptable risks in the event of a malfunction.Market Competition Shifts from Hardware to Software and Verification SupportCurrently, there is a growing number of memory modules on the market that emphasize support for functional safety. Most solutions highlight their hardware design compliance with ASIL-B or ASIL-D requirements, including built-in error correction mechanisms (ECC), redundancy design, data backup, and power-off protection.However, truly effective solutions that help developers lower the verification threshold and quickly integrate into main systems are rare.In practice, even if the memory hardware itself meets ISO 26262 safety design requirements, the overall system cannot smoothly complete functional safety verification without corresponding software support and certification documentation. This is the pain point most module developers encounter when introducing safety memory:*Lack of directly integrable software components and APIs*Inability to provide Safety Requirements Documents (SRD) and software verification reports*Software does not comply with ISO 26262 Part 6 development process requirementsIn other words, in the battlefield of functional safety, hardware alone is not enough—comprehensive software integration and verification resources are also needed to truly solve the problem.Winbond W77Q / W77T Safety Solution: A Comprehensive, Compliant Secure Flash Memory SolutionTo address the above market needs, Winbond has launched the W77Q/T secure flash memory solution for automotive ECU and module applications, providing comprehensive support from hardware, software, to verification processes, helping customers quickly achieve ISO 26262 compliance.Differentiator 1: Software Package Certified to ISO 26262 Part 6We offer not just hardware, but a software package that has been fully validated through the functional safety development process. This software suite includes:*Secure memory access drivers (including ECC and redundancy mechanisms)*API interface documentation and sample code*Software Safety Requirements Specification (SSR), design documents (SDD), and test reports*Complete safety case documentationAll these materials are developed and validated according to ISO 26262-6 standards and can be directly referenced by system developers in their main system functional safety documentation. In practice, customers do not need to conduct their own functional safety analysis for the memory module; simply integrating our documentation and software supports the product verification process, greatly saving costs and time.Differentiator 2: Modular Hardware Design for New and Legacy SystemsTo address the characteristics of the module market, W77Q / W77T supports the following application features:*Drop-in replacement for existing modules: Can directly replace existing flash components without modifying the PCB or driver architecture, enabling seamless upgrades.*Support for scalable architectures: Modules can be selected for ASIL-C or ASIL-D levels, meeting different module application needs (e.g., power control vs. lighting control).*Automatic power-off backup and independent safety processing unit (Safety Island): Enhances the module's ability to maintain minimum safety functions in the event of unexpected situations.This dual support design for "legacy system upgrades + new system acceleration" is especially suitable for module suppliers undergoing functional safety transformation.One-Stop Compliance Support Opens New Channels for Module MarketsOverall, the W77Q/ W77T secure flash solution is not just a component, but a complete "functional safety compliance accelerator."Credit: CompanyFor module and ECU developers, such a solution not only reduces the burden on internal functional safety engineers but also enables rapid completion of customer-required functional safety reports, accelerating product adoption by OEMs. This truly achieves a transition from compliance to competitiveness.To learn more about Winbond's advanced security solutions, visit Winbond's website or contact Winbond directly, or download the latest Hardware Security White Paper.
Global Unichip Corp. (GUC), the Advanced ASIC Leader, today announced the successful tape-out of the industry-leading Universal Chiplet Interconnect Express (UCIe) PHY Face-Up IP on TSMC's N5 process for integration with the TSMC SoIC-X technology. Targeted for AI, HPC, xPU, and networking applications, the IP achieves a breakthrough 36Gbps performance with Adaptive Voltage Scaling (AVS), delivering 2× better power efficiency at the required data rate. This solution offers an industry-leading bandwidth density of 1.5TB/s per mm of die edge. The chip was assembled using TSMC's advanced SoIC-X and CoWoS (Chip-on-Wafer-on-Substrate) packaging technologies.Earlier this year, GUC demonstrated UCIe-32G silicon on the TSMC N3P process at the TSMC 2025 North America Technology Symposium. In 2024, GUC also taped out its UCIe LP (Low Power) solution featuring AVS on the TSMC N5 process, aimed at meeting the growing bandwidth demands of multi-die integration in AI, HPC, and networking applications. Leveraging extensive experience in 3D interface IP and SoIC design, GUC developed a Face-Up version of UCIe LP IP, enabling robust die-to-die interconnect for the bottom die of SoIC-X configurations. Looking ahead, GUC is actively developing UCIe 64G IP, with plans to tape out by the end of 2025, addressing the ever-growing need for higher bandwidth in next-generation chiplet-based systems.To reduce PHY power consumption, all GUC UCIe LP IPs incorporate Adaptive Voltage Scaling (AVS), optimizing supply voltage and driving strength to improve power efficiency by up to 2x. A training algorithm dynamically selects minimal voltage and drive strength to meet eye margin criteria, ensuring reliable operation under varying voltage and temperature conditions. Additionally, the IP integrates proteanTecs' I/O signal quality monitors, allowing real-time performance monitoring without the need for re-training or data transfer interruption.For easy integration, GUC has developed bridges for AXI, CXS, and CHI buses utilizing the UCIe Streaming Protocol. These bridges are optimized for high traffic density, low latency, and low power, with efficient end-to-end flow control, facilitating a seamless transition from single-chip NoC architectures to chiplet-based systems. They support Dynamic Voltage and Frequency Scaling (DVFS), enabling on-the-fly changes in digital supply voltage and bus frequency while maintaining uninterrupted data flow."With industry-leading UCIe solutions now available on the TSMC N5 and N3P technologies, we are excited to introduce the brand new UCIe Face-Up IP for SoIC-X, supporting 36Gbps with twice the power efficiency," said Aditya Raina, CMO of GUC. "We've built a complete, silicon-proven 2.5D/3D chiplet IP portfolio across TSMC's 7nm, 5nm, and 3nm process technologies. Combined with our expertise in design, package integration, electrical and thermal simulation, DFT, and production testing, we deliver a comprehensive solution that accelerates development cycles and product ramp-up for AI, HPC, xPU, and networking customers.""Our mission is to provide the fastest and lowest-power 2.5D/3D chiplet interface IPs, enabling a smooth transition from monolithic SoCs to modular chiplet architectures," added Igor Elkanovich, CTO of GUC. "The convergence of 2.5D and 3D packaging, leveraging HBM3/4, UCIe, and GLink-3D interfaces, paves the way for highly modular processors that exceed traditional reticle limits."GUC UCIe LP Face-up IP HighlightsTo learn more about GUC's UCIe IP portfolio and TSMC's CoWoS and SoIC total solution, please contact GUC sales representative via email (https://www.guc-asic.com/en/contact)
Over 100 business and technology leaders from across our customer and partner network gathered for the EdgeOpen Consortium APAC 2025 event, held at ADLINK headquarters in Taoyuan, Taiwan, on May 19-20, 2025.Credit: CompanyStephen Huang, ADLINK's President and Chief Operating Officer, officially commenced the event with a keynote titled "Empower AI at the Edge: Scale by Industrial Application." He emphasized the critical role of customer insights in shaping ADLINK's strategic direction.By staying close to customer needs, Stephen Huang explained, ADLINK has critical insight into key challenges, especially the need for faster deployment, greater system intelligence, and reduced complexity. These insights have shaped ADLINK's response to global uncertainties: rather than simply delivering hardware, ADLINK focuses on integrated solutions that add software value to unlock the full potential of AI at the edge.ADLINK transforms market risks into business momentumStephen Huang acknowledges the growing challenges in today's global landscape, including trade tariffs, geopolitical tensions, overcapacity, and currency fluctuations. But rather than viewing these as roadblocks, ADLINK sees them as catalysts for innovation and momentum. Outlined key initiatives in response to these challenges:*Supply Chain Resilience: In response to global disruptions, ADLINK has established dual-site manufacturing in Taiwan and China, ensuring agility and continuity. *Operational Efficiency: With an integrated supply and value chain, ADLINK has streamlined operations to adapt quickly and reduce costs. Innovative *Breakthroughs: ADLINK is driving deeper capabilities in its proven solutions, such as the OSM module, SMU tester, rugged panel PCs, and Soft Motion software, to support more complex and connected edge scenarios.Credit: CompanyOver the past decades, ADLINK has evolved by creating a robust lineup of products designed for edge AI applications. This strong foundation now supports a new phase of innovation, software-integrated solutions that meet the growing demand for intelligent, connected systems. Looking ahead, General Manager Stephen Huang emphasized ADLINK's focus on key technology innovation and its continued close collaboration with industry leaders to develop core technologies that shape the future jointly.Concluding his keynote, he shared a vision of collaboration: "I hope that not only ADLINK, but also our partners, will continue to support and co-create Edge AI solutions, so we can unlock more opportunities and drive real business growth together."Global technology leaders share edge AI strategiesFollowing the opening keynote by Stephen Huang, ADLINK's President and COO, global technology leaders took the stage to share their strategic insights on the future of edge AI. The partner keynote series began with Intel, which highlighted how open ecosystems and scalable platforms are accelerating innovation at the edge. MediaTek followed with its perspective on enabling the next generation of AIoT applications across a wide range of industries. AUO Display Plus showcased how advancements in green display technologies are reshaping visual experiences in both industrial and commercial environments. NXP explored the role of machine learning in transforming smart factory operations and strengthening secure edge connectivity. Phison introduced breakthroughs in high-performance and endurance flash storage that not only reduce GPU cost but also enhance the inference performance of private generative AI. SynaXG concluded the session with a forward-looking perspective on expanding AI from centralized infrastructure to mobile and autonomous systems, such as robotics, connected vehicles, and drones. These strategic contributions set the tone for a day centered on real-world edge AI innovation and collaborative momentum.Credit: CompanyEcosystem collaboration with leading solution partnersComplementing the presence of global technology leaders, and prominent channel partners from across the Asia-Pacific region, including Macnica Altima Company, HF Technology, RTC Technology, and PB Technology, shared market insights, success stories, and collaborative achievements with ADLINK. Their involvement highlighted the strength of regional partnerships in accelerating edge AI adoption.Several solution partners also took the stage to present how their technologies integrate with ADLINK hardware to deliver real-world solutions. These included AUO, Elma Electronic Taiwan, Phenikaa-X, Tallgeese AI, Hailo, CyberLink, Wind River, Delta Electronics, AUO Display Plus, and Euresys. Their presentations demonstrated the power of ecosystem collaboration in delivering reliable, high-performance edge AI solutions tailored to a wide range of vertical markets, including manufacturing, healthcare, transportation, and smart cities.Accelerating edge AI adoption with scalable solutionsThe exhibition featured an application-focused showcase area, highlighting real-world implementations of ADLINK's edge AI solutions. From modular hardware and ruggedized devices to intelligent systems, the demonstrations illustrated how innovation is translated into practical solutions that help customers address frontline challenges.In smart manufacturing, ADLINK showcased its Soft-landing solution, empowered by SoftMotion EtherCAT controllers, along with a high-speed AI AOI inspection solution, enabling precise, multi-axis motion control and enhancing real-time, reliable quality management. For intelligent vision applications, ADLINK demonstrated a plug-and-play AI OCR solution powered by Euresys Open eVision software and the NEON smart camera, which adapts easily to diverse lighting conditions and requires no pre-training. In the area of interactive AI, ADLINK collaborated with Avalanche Computing, OSENSE, and Phison to launch an edge AI solution supporting private, offline LLM deployment. This enables the fast and low-cost implementation of AI assistants, smart customer service, and on-premises knowledge retrieval, ensuring both data privacy and system flexibility.ADLINK also introduced advanced AI-based safety applications and smart display technologies. In partnership with TeamMax, it presented an AI-powered fire and smoke detection solution for real-time monitoring and early warning in industrial and public spaces. Collaborating with AUO Display Plus, ADLINK showcased an energy-efficient HiRaso Panel PC and passenger information display systems (PIDS), highlighting the synergy between edge computing and next-generation smart displays. Additionally, in collaboration with its partner EGK, ADLINK developed a solution to enhance the in-vehicle experience, further advancing intelligent transportation systems. These comprehensive solutions demonstrate ADLINK's ability to deliver deployable, cross-industry edge innovations that accelerate AI adoption and drive digital transformation across sectors.
At COMPUTEX 2025, ASUS showcased its comprehensive AI infrastructure solutions, targeting the computing needs of data centers, edge computing, and a wide range of intelligent applications. With a robust product portfolio, ASUS brings to life the NVIDIA AI Factory vision, illustrating how AI will be deeply embedded into enterprise IT infrastructure. This evolution continues to drive forward applications and services that shape the future of AI and high-performance computing.Comprehensive Showcases: From Hardware to Data Center ManagementASUS's booth centered around a diversified server product line and integrated hardware-software systems. These are paired with data center management and monitoring solutions, presenting a well-rounded, detailed exhibit. In particular, ASUS features purpose-built architectures aligned with the NVIDIA AI Factory concept, designed to optimize next-generation data center infrastructure.As future AI applications require increasingly massive token processing capabilities, computing demands grow significantly. ASUS's new design architecture focuses on boosting performance and AI processing power while simultaneously addressing the challenges of thermal management. The booth highlighted versatile cooling options, including air- and liquid-cooled cabinet systems, making this one of the key attractions for visitors.Complete Lineup: NVIDIA HGX, NVIDIA MGX, and NVIDIA Blackwell Architectures for Rapid AI Infrastructure GrowthAt the entrance of the exhibition space, ASUS showcased its AI POD full-rack system based on NVIDIA HGX systems, supporting NVIDIA Blackwell and Blackwell Ultra GPUs. Known for its scale-up capabilities in a single server, the HGX system is a critical solution for high-performance computing needs.Next comes the NVIDIA MGX-based solutions, utilizing PCIe to deliver scalable and flexible AI computing power. ASUS emphasizes its certified, reference-architecture-based product line, validated by NVIDIA for performance and reliability, helping customers build independent, tailored AI infrastructure suited for various use cases.This year's highlight is the latest NVIDIA Grace Blackwell architecture. ASUS presents its NVIDIA GB200 NVL72 rack solution, targeting high-end AI POD full-rack systems featuring the NVIDIA GB200 and GB300 platforms. This powerful setup includes 72 Blackwell Ultra GPUs and 32 NVIDIA Grace CPUs, forming a system with 18 compute trays and 9 switch trays. Notably, the entire rack is equipped with a full liquid cooling system, maximizing computing density and energy efficiency, showcasing the next generation of performance-driven design.Software Tools Empower Enterprise AI DevelopmentOn the software front, ASUS supports customers from the early planning and deployment of AI server racks through infrastructure management and monitoring. Solutions such as ACC and AIDC offer one-stop deployment and operations support, addressing the entire data center lifecycle.ASUS also presented its collaboration with NVIDIA in deploying Agentic AI systems, custom AI agents capable of reasoning, planning, and action-taking. The integration of NVIDIA AI Workbench enables development teams to flexibly build applications on local or remote GPUs, streamlining workflows from experimentation and prototyping to proof-of-concept. This delivers an accessible, robust toolkit for enterprise AI development.Additionally, inspired by ChatGPT, ASUS introduces AI Hub, a software solution that simplifies LLM (Large Language Model) usage for enterprise teams. Now part of ASUS's AI infrastructure portfolio, AI Hub provides pre-integrated tools, open-source software, and LLM resources. It also supports collaboration with AI model developers, cloud service providers, and system integrators, helping enterprises more easily build and scale their AI applications and accelerate their innovation journey.