Search Results

Displaying Results 151-175 of 358 "'HBM'"

Apr 08, 2026, 07:29 ET Semidynamics consigue una inversión para impulsar chips de inferencia de IA centrados en la memoria

Semidynamics es capaz de ofrecer una capacidad de memoria varias veces superior a la disponible en los sistemas de inferencia convencionales basados en HBM, lo que permite admitir modelos, cachés KV y contextos más grandes. Estas tres características posibilitan un mayor número de usuarios por rack, lo que

More news about: Semidynamics


Apr 08, 2026, 05:36 ET Semidynamics Secures a Strategic Investment to Advance Memory-Centric AI Inference Chips

capacity and data movement rather than raw compute. Semidynamics is capable of delivering multiples of the memory capacity available in conventional HBM-based inference systems, hence supporting larger models, larger KV-caches, and larger contexts. These three features enable more users per rack, directly

More news about: Semidynamics


Apr 08, 2026, 05:27 ET Semidynamics Secures a Strategic Investment to Advance Memory-Centric AI Inference Chips

capacity and data movement rather than raw compute. Semidynamics is capable of delivering multiples of the memory capacity available in conventional HBM-based inference systems, hence supporting larger models, larger KV-caches, and larger contexts. These three features enable more users per rack, directly

More news about: Semidynamics


Apr 07, 2026, 18:00 ET SEMI Reports Global Semiconductor Equipment Billings Reached $135 Billion in 2025, Up 15% Year-on-Year

back-end equipment segment also recorded strong growth in 2025. Test equipment billings surged 55% year-over-year as AI devices and high-bandwidth memory (HBM) increased performance requirements and test intensity, while assembly and packaging equipment sales rose 21% as adoption of advanced packaging technologies

More news about: SEMI


Apr 07, 2026, 18:00 ET SEMI Reports Global Semiconductor Equipment Billings Reached $135 Billion in 2025, Up 15% Year-on-Year

back-end equipment segment also recorded strong growth in 2025. Test equipment billings surged 55% year-over-year as AI devices and high-bandwidth memory (HBM) increased performance requirements and test intensity, while assembly and packaging equipment sales rose 21% as adoption of advanced packaging technologies

More news about: SEMI


Apr 06, 2026, 10:15 ET Data Center Chip Market worth $687.65 billion by 2032 - Exclusive Report by MarketsandMarkets™

more memory-focused, performance relies more on memory bandwidth than solely on processing power, fueling the demand for advanced solutions such as HBM and DDR. Furthermore, the growing deployment of AI training and inference models, along with the need to handle massive datasets in parallel computing

More news about: MarketsandMarkets


Apr 02, 2026, 07:00 ET Roundhill Investments Launches First-Ever Memory ETF (DRAM)

exposure to global companies operating within the memory segment of the semiconductor industry, including manufacturers of DRAM, High-Bandwidth Memory (HBM), NAND flash, and solid-state storage devices (SSD). Memory semiconductors have emerged as a critical and increasingly constrained resource within the

More news about: Roundhill Investments


Apr 01, 2026, 10:00 ET SEMI Projects Double-Digit Growth in Global 300mm Fab Equipment Spending for 2026 and 2027

demand for memory has significantly increased due to AI training and inference. AI training has notably driven up the demand for High Bandwidth Memory (HBM), while model inference has created substantial demand for storage capacity, thus boosting the growth of NAND Flash applications in data centers. This

More news about: SEMI


Apr 01, 2026, 10:00 ET SEMI Projects Double-Digit Growth in Global 300mm Fab Equipment Spending for 2026 and 2027

demand for memory has significantly increased due to AI training and inference. AI training has notably driven up the demand for High Bandwidth Memory (HBM), while model inference has created substantial demand for storage capacity, thus boosting the growth of NAND Flash applications in data centers. This

More news about: SEMI


Mar 31, 2026, 03:44 ET INFINITIX、Phison とともに IT WEEK JAPAN にて異種 AI インフラソリューションを展示

学習を実現AI学習において企業が直面するGPUメモリ容量の制約および高コスト問題に対し、本ソリューションではPhisonのaiDAPTIV+技術を導入し、高速AI SSDを用いたGPUメモリの動的拡張を実現する。ストレージ層をAI演算アーキテクチャに組み込むことで、高価な HBM GPUを全面的に導入することなく、大規模モデルの学習を可能にする。これにより、計算資源利用効率と投資対効果を向上させ、企業に対してより柔軟なインフラ構成戦略を提供する。インテリジェントオーケストレーションにより効率を向上、TCO 最適化を実現

More news about: INFINITIX Inc.


Mar 30, 2026, 09:53 ET Harbour BioMed Reports Full Year 2025 Financial Results: Sustained Global Collaboration Underpins Long-Term Growth

commercialization of HBM9378 globally (excluding Greater China and several Southeast and West Asian countries). In July 2025, Windward Bio launched global Phase II POLARIS clinical study, assessing long-acting dosing of HBM9378/WIN378 for people living with asthma.Note: HBM9378 is known as SKB378

More news about: Harbour BioMed


Mar 30, 2026, 08:10 ET CAMTEK RECEIVES $31 MILLION MULTI-SYSTEM ORDER FROM A LEADING OSAT

inspect wafers for the most demanding semiconductor market segments, including Advanced Interconnect Packaging, Heterogenous Integration, Memory and HBM, CMOS Image Sensors, Compound Semiconductors, MEMS, and RF, serving numerous industry's leading global IDMs, OSATs, and foundries.With manufacturing

More news about: Camtek Ltd.


Mar 24, 2026, 01:00 ET Kulicke & Soffa Expands Memory Solutions Portfolio

customers and technology partners to address this potential industry opportunity.As emerging memory architectures such as high–bandwidth memory (HBM), high–bandwidth flash (HBF), and other forms of high-density DRAM become more prevalent, K&S FTC and Hybrid solutions are well positioned to provide

More news about: Kulicke & Soffa Industries, Inc.


Mar 22, 2026, 23:14 ET Harbour BioMed Reports the Online Publication of Phase I Results for HBM9378 (SKB378/WIN378), a TSLP-Targeting Antibody

evaluate the safety, tolerability, pharmacokinetics (PK) and immunogenicity of HBM9378. The dose-escalation study enrolled ten healthy subjects per cohort (20 mg, 60 mg, 200 mg, 600 mg, and 900 mg), with eight receiving HBM9378 and two receiving the placebo at each dose level.Data showed that the

More news about: Harbour BioMed


Mar 20, 2026, 00:31 ET 슈퍼마이크로, DCBBS에 차세대 엔비디아 베라 루빈 및 루빈 플랫폼 탑재

루빈 GPU, 베라 CPU, NVLink 6, 커넥트 X-9 슈퍼 NIC, 블루필드-4 DPU, 스펙트럼-X 이더넷 등 6개 핵심 칩을 통합한다.최대 3.6 엑사플롭스의 추론 성능, 75TB 고속 메모리, 최대 1.6PB/s HBM4 대역폭을 제공하며, 이전 세대 대비 와트당 처리량을 최대 10배 향상시키고 토큰 비용을 10분의 1 수준으로 낮추는 것을 목표로 한다.또한 엔비디아 MGX 랙 아키텍처 기반으로 설계되어 랙 및 클러스터 규모 AI 인프라의 효율적인

More news about: Super Micro Computer, Inc.


Mar 18, 2026, 08:05 ET 基於 NVIDIA BlueField-4 STX,WEKA 以更低的每 Token 成本將 Token 輸出推至極限

利用共享鍵值 (KV) 緩存基礎設施解決推理成本問題擴展代理型系統,尤其在軟件工程應用領域,揭示一個殘酷真相:現今 AI 的經濟效益取決於記憶體基礎設施層。 每個大規模推理集群都會撞上記憶體牆:GPU 上有限的高頻寬記憶體 (HBM) 很快便耗盡,導致 KV 快取被逐出、情境丟失,系統被迫重複已經完成的工作。 這種架構效率低下,令推理成本急升。 解決之道在於建立共享的 KV 快取基礎設施,讓情境在代理、用戶與會話之間保持活躍。 這樣能消除重複計算、維持 Token 吞吐量,並保持效能穩定可測。 缺乏共享 KV 快取基礎設施的話,每增加一批並發用戶及代理,都會成為負累——成本上漲、體驗變差,推理集群規模越大,營運就越難。 NVIDIA

More news about: WEKA


Mar 18, 2026, 01:25 ET Supermicro onthult DCBBS® met nieuwe NVIDIA Vera Rubin NVL72-, HGX Rubin NVL8- en Vera CPU-systemen, ontworpen om de marktintroductietijd van klanten te versnellen

6, NVIDIA ConnectX-9 SuperNIC, NVIDIA BlueField-4 DPU en NVIDIA Spectrum-X Ethernet - om tot 3,6 Exaflops inferentie, 75 TB snel geheugen en 1,6 PB/s HBM4-bandbreedte te leveren, met als doelstelling tot 10x de doorvoer per watt en een tiende van de tokenkosten in vergelijking met NVIDIA Blackwell.NVIDIA

More news about: Super Micro Computer, Inc.


Mar 17, 2026, 18:43 ET WEKA, NVIDIA BlueField-4 STX पर प्रति टोकन कम लागत के साथ टोकन आउटपुट को अधिक से अधिक बढ़ाता है

मेमोरी इंफ़्रास्ट्रक्चर लेयर पर होता है। हर बड़े पैमाने पर अनुमान लगाने वाले बेड़े की मेमोरी सीमा आ जाती है: GPU पर मौजूद सीमित हाई-बैंडविड्थ मेमोरी (HBM) जल्दी खत्म हो जाती है, की-वैल्यू (KV) कैश हट जाता है, कंटेक्स्ट खो जाता है, और सिस्टम को वही काम दोबारा करना पड़ता है जो वह पहले कर चुका होता है। इस

More news about: WEKA


Mar 17, 2026, 18:40 ET WEKA optimise la production de jetons avec un coût inférieur par jeton sur NVIDIA BlueField-4 STX

l'infrastructure de la mémoire. Chaque flotte d'inférence à grande échelle se heurte au mur de la mémoire : la mémoire limitée à grande largeur de bande (HBM) du GPU est rapidement épuisée, le cache clé-valeur (KV) est évincé, le contexte est perdu et le système est contraint de répéter le travail qu'il a

More news about: WEKA


Mar 17, 2026, 08:00 ET China Controls the Metal Underlying America's Trillion-Dollar Tech Economy - OilPrice.com Market Commentary

because their next-generation memory, HBM4, is the secret sauce that makes AI accelerators work. The production of these stacked-die chips uses rare earth dopants to maintain signal integrity at extreme speeds, and Micron has already sold out its entire 2026 HBM4 capacity under multi-year contracts.The

More news about: OilPrice.com


Mar 17, 2026, 07:19 ET A Supermicro revela o DCBBS® com os novos sistemas NVIDIA Vera Rubin NVL72, HGX Rubin NVL8 e CPU Vera, projetados para acelerar o tempo de lançamento no mercado de seus clientes

BlueField-4 DPU e NVIDIA Spectrum-X Ethernet — para fornecer até 3,6 Exaflops de inferência, 75 TB de memória rápida e 1,6 PB/s de largura de banda HBM4, visando até 10 vezes mais throughput por watt e um décimo do custo do token em comparação com o NVIDIA Blackwell.Sistema NVIDIA HGX Rubin

More news about: Super Micro Computer, Inc.


Mar 16, 2026, 16:37 ET SK hynix to Showcase AI Memory Leadership at NVIDIA GTC 2026

GPU-based AI accelerators through physical models and actual hardware, focusing on the actual application of SK hynix's memory products such as HBM4, HBM3E, and SOCAMM2 designed for NVIDIA AI platforms.Notably, the company will showcase a liquid-cooled eSSD developed in collaboration

More news about: SK hynix Inc.


Mar 16, 2026, 16:35 ET Supermicro Reveals DCBBS® with New NVIDIA Vera Rubin NVL72, HGX Rubin NVL8, and Vera CPU Systems, Designed to Accelerate Customer Time-to-Market

SuperNIC, NVIDIA BlueField-4 DPU, and NVIDIA Spectrum-X Ethernet — to deliver up to 3.6 Exaflops of inference, 75TB of fast memory, and 1.6 PB/s of HBM4 bandwidth, targeting up to 10x the throughput per watt and one-tenth the token cost compared to NVIDIA Blackwell.NVIDIA HGX Rubin

More news about: Super Micro Computer, Inc.


Mar 16, 2026, 16:35 ET Supermicro Reveals DCBBS® with New NVIDIA Vera Rubin NVL72, HGX Rubin NVL8, and Vera CPU Systems, Designed to Accelerate Customer Time-to-Market

SuperNIC, NVIDIA BlueField-4 DPU, and NVIDIA Spectrum-X Ethernet — to deliver up to 3.6 Exaflops of inference, 75TB of fast memory, and 1.6 PB/s of HBM4 bandwidth, targeting up to 10x the throughput per watt and one-tenth the token cost compared to NVIDIA Blackwell.NVIDIA HGX Rubin NVL8 SystemThe

More news about: Super Micro Computer, Inc.


Making a selection with these dropdown will cause content on this page to change. News search result will update as each option is selected.