Connect with us

Technology

MiTAC Computing Showcases Diversified Infra for Agentic Workloads at OCP APAC Summit

Published

on

TAIPEI, Aug. 12, 2026 /PRNewswire/ — MiTAC Computing Technology Corporation, a global leader in high-performance, energy-efficient server solutions and a subsidiary of MiTAC Holdings Corporation (TSE:3706), today showcased its diversified rack-scale infrastructure engineered for Agentic AI workloads at the OCP APAC Summit. Highlighting the event is the world premiere of MiTAC’s latest OCP-compliant HPC servers—the C2810Z6 and C2610Z6—developed in collaboration with AMD to harness the all-new 6th Gen AMD EPYC™ Server CPUs on the advanced SP7/SP8 platforms with PCIe Gen 6 connectivity.

To meet the demands of continuous Multi-Agent Reasoning Loops and extreme data throughput, MiTAC has engineered comprehensive rack-scale solutions centered around its flagship 52U high-density AI liquid-cooled rack. By incorporating integrated AI servers that span AMD and Intel platforms alongside the NVIDIA accelerated computing platform—featuring compute, networking, and software—MiTAC provides a seamless upgrade path from underlying silicon to a ready-to-deploy ‘AI Factory’.

Premiering Next-Gen OCP Compute Nodes for Agentic Execution

As autonomous agents transition from simple prompts to dynamic multi-step workflows, foundational compute nodes require unprecedented I/O bandwidth and power efficiency. At OCP APAC, MiTAC is debuting two ORv3 compute platforms powered by 6th Gen AMD EPYC™ Server CPUs. Featuring AMD’s “Zen 6” architecture, PCIe Gen 6 connectivity, and CXL 3.1 support, these platforms are engineered to process massive parallel agent requests while optimizing data center footprint and energy density.

MiTAC leads this architectural transition with two world-premiere platforms:

C2810Z6: An ORv3 multi-node server engineered for high-density cloud and HPC workloads. Powered by 6th Gen AMD EPYC™ Server CPUs (SP7 socket, up to 256 cores and 512 threads,supporting default CPU power up to 600W), it supports up to 4TB DDR5-8000 memory per node. This platform delivers massive PCIe Gen 6 I/O throughput alongside versatile NVMe storage (12x E3.S-1T or 6x U.2), such as Micron 9650 NVMe™ Gen 6 SSDs. Integrated with a 48Vdc busbar and DC-SCM 2.0 module, it ensures extreme performance, security, and power efficiency for dense compute clusters.C2610Z6: A next-generation ORv3 compute platform optimized for scalable cloud, enterprise IT, and edge applications. Powered by 6th Gen AMD EPYC™ Server CPUs (SP8 socket, up to 128 cores and 256 threads, supporting default CPU power up to 400W), it supports up to 4TB DDR5-8000 memory per node. Equipped with PCIe Gen 6 connectivity, flexible NVMe storage, such as Micron 9650 NVMe™ Gen 6 SSDs. Integrated with a 48Vdc busbar and DC-SCM 2.0 management, it provides a highly efficient, versatile architecture for enterprise AI scaling.

Diversified Rack-Scale Infrastructure for Agentic Workloads

Beyond raw node performance, powering autonomous Agentic AI introduces complex architectural demands. Unlike static single-prompt AI, multi-agent systems rely on continuous reasoning loops, dynamic tool calling, real-time memory retrieval, and multi-agent orchestration.

To eliminate thermal, storage, and cluster governance bottlenecks across these execution pipelines, MiTAC Computing delivers a complete spectrum of diversified rack-scale infrastructure:

High-Density AI Racks (Liquid-Cooled & Air-Cooled): Engineered to sustain the continuous multi-step reasoning loops of agentic workloads. By packing maximum GPU density—such as 96x AMD Instinct™ MI355X GPUs in a single 52U liquid-cooled rack—MiTAC provides the low-latency, unthrottled compute engine required for real-time agent execution.OCP HPC Racks (Modular & Scalable): Delivering open ORv3 standards, energy-efficient 48Vdc power distribution, and modular flexibility, MiTAC’s liquid- and air-cooled OCP racks allow data centers to scale execution nodes seamlessly as autonomous agent worker teams expand.Agentic Memory & Orchestration Ecosystem (DDN, Rafay, and Canonical Ubuntu): An autonomous agent is only as fast as its access to enterprise memory and task delegation. Partnering with DDN for ultra-high-throughput vector/session storage and Rafay for automated multi-agent cluster governance, MiTAC delivers an ultra-low latency storage-to-compute pipeline that keeps multi-agent workflows running at full speed.

From Chips to Applications: MiTAC’s Full-Stack AI Factory Vision

To bring these diversified components together into an operational environment, MiTAC aligns its entire rack-scale portfolio across ecosystem layers to transform modular silicon into deployment-ready enterprise AI factories:

MiTAC platforms maximize the raw compute of next-generation silicon, offering the flexibility to integrate high-density accelerators—including AMD Instinct™ MI355X GPUs, AMD Instinct™ MI350X GPUs, AMD Instinct™ MI350P PCIe® cards—as well as multi-workload enterprise GPUs such as the NVIDIA RTX PRO 4500 and RTX PRO 6000 Blackwell Server Edition GPUs, ideally suited for AI, simulation, and visual computing.Infrastructure Layer: We deliver hyper-efficient physical foundations—highlighted by our 52U AI liquid-cooled rack housing G4826Z5 servers alongside versatile compute nodes like the TN85-B8261 and G4520G6 (powered by dual Intel® Xeon® 6 processors with up to 8,192GB DDR5 memory)—backed by MiOBMC™ / MiOPF™ open firmware and enterprise Linux OS.Orchestration & Data Pipeline Layer: Bridging physical servers with upper-layer intelligence, MiTAC collaborates with leading ecosystem partners including DDN, Rafay, and Kubernetes, alongside our proprietary MiCoreView™ POD management suite, to deliver high-throughput data delivery and automated cluster management across modern AI factories.Models Layer: Our throughput-optimized compute and parallel storage pipelines satisfy the demands of diverse architectures—accelerating massive LLMs, VLMs, MMLLMs, and Mixture-of-Experts (MoE) models without data bottlenecks or hardware idling.Applications Layer: From multi-step Agentic AI and RAG workflows to generative AI and large-scale HPC simulations, MiTAC provides resilient, rack-scale foundations purpose-built to execute real-world enterprise AI at scale.

About MiTAC Computing Technology Corporation

MiTAC Computing Technology Corp., a subsidiary of MiTAC Holdings, delivers comprehensive, energy-efficient server solutions backed by industry expertise dating back to the 1990s. Specializing in AI, HPC, cloud, and edge computing, MiTAC Computing employs rigorous methodologies to ensure uncompromising quality—across barebones, systems, racks, and cluster levels—fully achieving performance and integration. This commitment to quality at every level sets MiTAC Computing apart in the industry.

With a worldwide presence and end-to-end capabilities—from R&D and manufacturing to global support—MiTAC Computing provides agile, customized platforms for hyperscale data centers, HPC, and AI applications, ensuring optimal performance and scalability to meet unique business needs. By leveraging the latest advancements in AI and liquid cooling, and unifying the MiTAC brand with Intel DSG and TYAN server products, MiTAC Computing stands out for its innovative, efficient, and reliable server technology as well as its hardware and software integrated solutions—empowering businesses to meet future challenges.

MiTAC Computing Technology Corporation website: https://www.mitaccomputing.com/.

View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/mitac-computing-showcases-diversified-infra-for-agentic-workloads-at-ocp-apac-summit-302849469.html

SOURCE MiTAC Computing Technology Corporation

Continue Reading

Technology

AtkinsRéalis Announces Private Offering of $400 Million of Unsecured Debentures

Published

on

By

MONTREAL, Nov. 9, 2026 /CNW/ — AtkinsRéalis Group Inc. (TSX: ATRL), a world-class engineering services and nuclear company with offices around the world, announced today that it has priced an offering of $400 million aggregate principal amount of 4.062% Series 11 unsecured debentures due November 12, 2029 (the “Debentures”).

Offering of Debentures

The Debentures are being offered through a syndicate of agents led by CIBC Capital Markets, Scotiabank and TD Securities, as joint bookrunners, and which also includes BMO Capital Markets, BNP Paribas (Canada) Securities Inc., National Bank of Canada Capital Markets, RBC Capital Markets, Desjardins Capital Markets and BofA Securities. The Debentures will be issued at a price of $1,000 per $1,000 principal amount of Debentures. AtkinsRéalis intends to use the net proceeds of the offering to repay existing indebtedness and for general corporate purposes.

The Debentures will bear interest at a fixed annual rate of 4.062%, payable in equal semi-annual instalments, in arrears, on May 12 and November 12 of each year until maturity, commencing on May 12, 2027.

The Debentures will be direct unsecured obligations of AtkinsRéalis, will rank pari passu with all other unsecured and unsubordinated indebtedness of AtkinsRéalis (including all of its other existing debentures as well as indebtedness under its existing main credit agreement) and will be guaranteed by the same subsidiaries that guarantee AtkinsRéalis’ obligations under its existing main credit agreement and its other outstanding debentures.

The Debentures have been assigned a provisional rating of BBB with a stable trend by DBRS Limited (Morningstar DBRS) and BBB- by Standard & Poor’s.

The Debentures have not been and will not be qualified for sale to the public under applicable securities laws in Canada and, accordingly, any offer and sale of the Debentures in Canada will be made on a basis which is exempt from the prospectus requirements of such securities laws. No securities regulatory authority has either approved or disapproved of the contents of this press release. The Debentures have not been and will not be registered under the United States Securities Act of 1933, as amended (the “U.S. Securities Act”), or the securities laws of any other jurisdiction, and may not be offered or sold in the United States absent registration under, or an applicable exemption from the registration requirements of, the U.S. Securities Act. This press release shall not constitute an offer to sell or the solicitation of an offer to buy, nor shall there be any offer to sell or a solicitation of an offer to buy the Debentures in any jurisdiction where it is unlawful to do so.

About AtkinsRéalis

Created by the integration of long-standing organizations dating back to 1911, AtkinsRéalis is a world-class engineering services and nuclear company dedicated to engineering a better future for our planet and its people. We create sustainable solutions that connect people, data and technology to transform the world’s infrastructure and energy systems. We deploy global capabilities locally to our clients and deliver unique end-to-end services across the whole life cycle of an asset including consulting, advisory & environmental services, intelligent networks & cybersecurity, design & engineering, procurement, project & construction management, operations & maintenance, decommissioning and capital advisory services. The breadth and depth of our capabilities are delivered to clients in strategic sectors such as Engineering Services and Nuclear, as steward of CANDU® nuclear technology. News and information are available at www.atkinsrealis.com or follow us on LinkedIn.  

Forward-Looking Statements
References in this press release, and hereafter, to the “Company”, “AtkinsRéalis”, “we”, “us” and “our” mean, as the context may require, AtkinsRéalis Group Inc. and all or some of its subsidiaries or joint arrangements or associates, or AtkinsRéalis Group Inc. or one or more of its subsidiaries or joint arrangements or associates.

Statements made in this press release, including, but not limited to, statements relating to the offering of the Debentures, or any other future events or developments and other statements that are not historical facts, constitute “forward-looking statements” which can be identified by the use of the conditional or forward-looking terminology such as “aims”, “anticipates”, “assumes”, “believes”, “cost savings”, “estimates”, “expects”, “forecasts”, “goal”, “intends”, “likely”, “may”, “objective”, “outlook”, “plans”, “projects”, “should”, “synergies”, “target”, “vision”, “will”, or the negative thereof or other variations thereon.

Specific forward-looking statements in this press release include, but are not limited to, statements with respect to the offering of the Debentures and the intended timing and use of proceeds thereof and to the credit ratings assigned to the Debentures. All such forward-looking statements are made pursuant to the “safe-harbour” provisions of applicable Canadian securities laws. The Company cautions that, by their nature, forward-looking statements involve risks and uncertainties, and that its actual actions and/or results could differ materially from those expressed or implied in such forward-looking statements, or could affect the extent to which a particular projection materializes. Forward-looking statements are presented for the purpose of assisting investors and others in understanding certain key elements of the Company’s current objectives, strategic priorities, expectations and plans, and in obtaining a better understanding of the Company’s business and anticipated operating environment. Readers are cautioned that such information may not be appropriate for other purposes.

Forward-looking statements made in this press release are based on a number of assumptions believed by the Company to be reasonable as at the date of this press release. The assumptions are set out throughout the Company’s 2025 annual management’s discussion and analysis (“2025 MD&A”) (particularly in the sections entitled “Critical Accounting Judgements and Key Sources of Estimation Uncertainty” and “How We Analyze and Report Our Results”) filed with the securities regulatory authorities in Canada, available on SEDAR+ at www.sedarplus.ca and on the Company’s website at www.atkinsrealis.com  under the “Investors” section. If these assumptions are inaccurate, the Company’s actual results could differ materially from those expressed or implied in such forward-looking statements. In addition, important risk factors could cause the Company’s assumptions and estimates to be inaccurate and actual results or events to differ materially from those expressed in or implied by these forward-looking statements. Those risks are identified in the 2025 MD&A (particularly in the section entitled “Risks and Uncertainties”) and may be updated from time to time in the Company’s 2026 interim quarterly management’s discussion and analysis.

The forward-looking statements herein reflect the Company’s expectations as at the date of this press release and are subject to change after this date. The Company does not undertake to update publicly or to revise any written or oral forward-looking information or statements, whether as a result of new information, future events or otherwise, unless required by applicable legislation or regulation. The forward-looking information and statements contained herein are expressly qualified in their entirety by this cautionary statement.

For More Information :

Media             

Investors   

Antoine Calendrier           

Denis Jasmin   

Vice President, Global External 

Vice President, Investor Relations   

Communications

514-393-8000, ext. 57553  

media@atkinsrealis.com 

denis.jasmin@atkinsrealis.com 

SOURCE AtkinsRéalis

Continue Reading

Technology

Korea Deep Learning Cuts Financial Document Processing Time by 88%, Targets APAC Financial Sector

Published

on

By

VLM-based AI agent demonstrates potential in automating complex financial workflows through joint PoC with South Korea’s Industrial Bank of KoreaKorean AI company advances PoC discussions with nine Japanese financial institutions following projects in Vietnam and Taiwan region

SEOUL, South Korea, Nov. 10, 2026 /PRNewswire/ — Korea Deep Learning (KDL), an enterprise AI agent company, is expanding its presence in the Asia-Pacific financial sector following a successful proof of concept (PoC) with South Korea’s Industrial Bank of Korea (IBK), which demonstrated an 88% reduction in financial document processing time.

The project applied KDL’s VLM-based AI agent, DEEP Agent for Finance, to automate document-intensive procedures for handling voice phishing relief cases. The solution combines document recognition, information extraction and crime-type inference, addressing tasks that traditionally require substantial manual review by financial institution employees.

During the PoC, the system achieved 100% accuracy in printed-text recognition and document classification, while handwriting recognition accuracy exceeded 97%, based on the project’s evaluation criteria. All key performance indicators exceeded their targets using actual business data. The solution was deployed within IBK’s closed network, with employee review retained for cases requiring additional verification.

The results provide a reference case for KDL as it pursues opportunities with financial institutions across Asia. The company has completed a separate PoC with Shinhan Vietnam Finance Company (SVFC), a local subsidiary of Shinhan Card, through the Shinhan Financial Group’s open innovation program. KDL has also supplied AI solutions in overseas markets including Vietnam and Taiwan region.

In Japan, KDL is advancing discussions about potential PoCs with nine financial institutions. It is also engaging with financial institutions across the region in banking, insurance and consumer finance. Some prospective partners have signed non-disclosure agreements and are evaluating the technology using actual business data.

KDL plans to convert ongoing technical evaluations into paid PoCs and supply contracts, while developing local partnerships and sales channels from 2027. The company also aims to extend its solutions beyond banking into insurance and securities, where document processing and information extraction remain important parts of daily operations.

“Financial institutions handling voice phishing relief cases often spend considerable time processing documents before they can focus on supporting customers,” said Ji-hyun Kim, CEO of Korea Deep Learning. “Our project with IBK demonstrates how VLM-based AI agents can reduce that administrative burden. We aim to help financial institutions automate repetitive tasks, improve operational efficiency and allocate more resources to consumer protection.”

About Korea Deep Learning

Founded in 2019, Korea Deep Learning is an enterprise AI agent company specializing in document intelligence and workflow automation. Its DEEP Agent platform combines DEEP OCR and DEEP Parser technologies to extract and process information from complex documents. The company serves more than 80 companies and public institutions in South Korea.

View original content to download multimedia:https://www.prnewswire.com/apac/news-releases/korea-deep-learning-cuts-financial-document-processing-time-by-88-targets-apac-financial-sector-302931894.html

SOURCE Korea Deep Learning

Continue Reading

Technology

LG Innotek Shines Again in Vehicle Lighting with Third Consecutive CES Innovation Award

Published

on

By

Next-generation vehicle lighting module Nexlide C+ wins CES 2027 Innovation AwardDifferent light patterns appear depending on the viewing angle, even from a single lighting moduleWorld’s first use of optical films in an ultra-slim surface light source module, enabling a thinner design

SEOUL, South Korea, Nov. 9, 2026 /PRNewswire/ — LG Innotek (CEO Moon Hyuksoo, 011070) announced on November 9 that its next-generation vehicle lighting module, Nexlide C+, has received a CES (Consumer Electronics Show) 2027 Innovation Award.

With its Nexlide series, LG Innotek has now received CES Innovation Awards for three consecutive years, from 2025 through 2027. The latest recognition reaffirms the company’s world-class competitiveness in vehicle lighting technology.

Nexlide is LG Innotek’s brand of next-generation flexible vehicle lighting modules, which combine the company’s proprietary optical design with advanced materials technology. Built on surface light source technology, which emits light in a broad and uniform fashion, Nexlide modules are thinner and more compact than conventional vehicle lighting while allowing a wide range of design possibilities. LG Innotek holds more than 700 patents related to surface light source technology.

The key strength of the award-winning Nexlide C+ is that a single lighting module can produce two or more 3D images, known as the 3D Multi-Effect. For example, the same module displays a 3D image with a diagonal pattern when viewed from the left and a vertical pattern when viewed from the right. Because the image changes with the viewing angle, the lighting creates a dynamic visual effect with a sense of depth.

Nexlide C+ can be integrated into various parts of a vehicle, from exterior lighting such as taillights and front grille lights to interior ambient lighting.

An LG Innotek official said, “Vehicle lighting is evolving from a simple signaling function into a key factor that shapes the design competitiveness of an automaker’s brand. Lighting technologies like Nexlide C+ that are capable of creating diverse light patterns will see even wider use.”

Nexlide C+ is also ultra-slim, at just 3 mm (0.12 in) thick or less, allowing it to be installed in tight spaces within a vehicle. LG Innotek is the first in the world to apply Light Turning Film and 3D Micro-Optic Pattern Film to an ultra-slim surface light source module.

The Light Turning Film precisely controls the direction of light to make different images appear at different viewing angles, while the 3D Micro-Optic Pattern Film makes images look three-dimensional even on a flat lighting surface.

The two films are only about 0.16 mm (0.0063 in) and 0.13 mm (0.0051 in) thick, respectively. By incorporating these ultra-thin films into the lighting module, LG Innotek enables a variety of 3D lighting effects in a slim profile. Previously, achieving the same effects required optical components such as plastic lenses and reflectors, and installing multiple components made the lighting unit bulkier.

By using these films to reduce the thickness of the lighting module, Nexlide C+ gives automakers even greater design freedom.

Going forward, LG Innotek will actively promote Nexlide C+ to global automakers and parts suppliers, targeting a 2027 launch.

At CES 2027 in January, LG Innotek will showcase a broad lineup of next-generation vehicle lighting modules, including Nexlide C+, along with key components for various areas of physical AI, such as autonomous driving and robotics.

[Background] LG Innotek previously received a CES 2025 Innovation Award for Nexlide A+, the industry’s first vehicle front lighting module to apply surface light source technology, with 40% less thickness and more than five times the brightness of the previous product. At CES 2026, the company received another Innovation Award for Nexlide Pixel, which reduced the lighting pixel size to among the smallest in the world while improving luminous efficiency by 30% compared with the previous product.

View original content to download multimedia:https://www.prnewswire.com/news-releases/lg-innotek-shines-again-in-vehicle-lighting-with-third-consecutive-ces-innovation-award-302933258.html

SOURCE LG Innotek

Continue Reading

Trending