Today's Hardware Pulse
Nvidia and SK Hynix ink multi-year AI memory chip pact
Nvidia and SK Hynix have signed a multi‑year agreement to co‑develop next‑generation memory chips optimized for artificial‑intelligence workloads. The collaboration will focus on memory for Nvidia’s Vera Rubin accelerator and broader AI infrastructure, pairing Nvidia’s AI expertise with SK Hynix’s high‑performance memory capabilities.
Also developing:
By the numbers: Rapidus raises $943M in funding
Microchip Timing Module Supports AI Data Centre and 5G Synchronization
Microchip has launched the MD-990-0011-B plug‑in timing module, co‑developed with Intel for Xeon 6 SoC servers and virtualized RAN platforms. The module offers automatic source selection among GNSS, SyncE and PTP, delivering resilient synchronization for AI workloads, cloud services and 5G radio access networks. Two variants provide 4‑hour or 8‑hour holdover to maintain timing during signal loss. By eliminating custom timing circuitry, the solution speeds up product development and reduces time‑to‑market.
GPUs: A High-Throughput Architecture Confronting a Workload Shift
Graphics processing units still power large‑scale AI training, but frontier large language models are exposing a growing memory bottleneck. While Nvidia H100 GPUs can deliver petaflops of FP8 throughput, trillion‑parameter inference often falls below 10 FLOPs per byte, making bandwidth...
Tattu Launches 5.0 Smart Battery Platform as Heavy-Lift Drones Demand Faster Charging and Smarter Fleet Power Management
Tattu, the UAV battery arm of Shenzhen Grepow, unveiled its 5.0 Smart Battery Platform aimed at 100 kg‑class agricultural and logistics drones. The system combines 6C fast charging, 12C continuous discharge, an intelligent BMS, and lifecycle data access, and pairs with...
Analog Devices Eyes $1.5 Bn Cash Acquisition of AI Chipmaker Empower Semiconductor
Analog Devices, Inc. is in advanced negotiations to purchase AI‑chip specialist Empower Semiconductor for roughly $1.5 bn in cash. The deal would add Empower’s FinFast power‑management technology to ADI’s portfolio as data‑center AI spending accelerates.
Escarda Starts Series Production of AI Laser Weeder After Acquisition
Berlin‑based Escarda Technologies, now fully owned by Berlin Industrial Group (B.I.G.), is moving into series production of its AI‑powered laser weeder. The system blends high‑resolution cameras, AI plant recognition and Class‑1 laser technology to eliminate weeds in real time without...
The Next 15 Years of Moore’s Law, According to Imec
Imec’s new 15‑year roadmap predicts the commercial debut of complementary FET (CFET) technology around 2033, effectively stacking PMOS and NMOS devices to halve circuit area. The institute also foresees a shift to two‑dimensional semiconductor channels by 2041 to boost power...
ChargePoint Wants To Fix The Condo EV Charging Problem With 2,500 New Ports
ChargePoint announced a plan to deploy roughly 2,500 Level 2 EV charging ports at multifamily residences across the United States, beginning in 2026. The rollout is backed by a partnership with OBE Power, which will own, operate, and handle all maintenance...
Blackstone and Google Launch $5 B AI Cloud Venture to Deploy Custom TPUs
Blackstone and Google announced a $5 billion joint venture to create a U.S.-based AI cloud company that will offer Google’s custom Tensor Processing Units as a compute‑as‑a‑service. The venture aims to bring 500 MW of data‑center capacity online in 2027, signaling deep...
Meta’s $200 B AI Data Center Redraws Rural Louisiana Real‑Estate Map
Meta Platforms is pouring more than $200 billion into Hyperion, a 5‑gigawatt AI data center in Richland Parish, Louisiana. The project will host up to 7,500 construction workers and reshape a swath of farmland the size of Manhattan, prompting a dramatic...
Intel and Qualcomm Eye Tenstorrent in AI Chip Consolidation Push
AI accelerator startup Tenstorrent has entered early-stage acquisition talks with Intel and Qualcomm. The interest reflects mounting pressure on major silicon vendors to secure more efficient AI processing technology and to counter Nvidia's dominance.
Rapid Adoption of Automotive Ethernet Raises Standardization and Cost Concerns
Tech Insights forecasts automotive Ethernet sockets will climb from about 962,000 in 2025 to 3.42 million by 2032, pushing the average per‑vehicle ports from 11 today to 27 by 2030. The rapid growth is driven by higher‑bandwidth sensors, autonomous‑driving workloads and...
Linux 7.2 Will Fix The Intel P-State Driver For The New Bartlett Lake CPUs
The upcoming Linux 7.2 kernel will incorporate a patch that corrects the Intel P‑State driver’s handling of the new Bartlett Lake processors, eliminating bogus 7 GHz maximum frequency reports. The fix originated from a QNAP engineer after the Intel Core 9 273PE was mistakenly...
The Nvidia H200 China Deal Survived the Trump-Xi Summit–Just Not in the Way Anyone Expected
President Trump’s Beijing visit ended with no new Nvidia H200 shipments, despite the chips receiving U.S. export licences in December 2025. Ten Chinese firms—including Alibaba, Tencent, ByteDance and JD.com—have licences for up to 75,000 units each, but Beijing has barred...
Samsung Electronics Launches Emergency Semiconductor Production Plan Ahead of Union Strike
Samsung Electronics activated an emergency management plan for its semiconductor fabs ahead of a union strike slated for May 21. The company is trimming wafer input volumes and shifting capacity to high‑bandwidth memory (HBM) chips to protect critical AI‑related supply....
Memory Chip Prices Set to Ease in Late 2027 as AI Demand Drives Global Market Volatility
Memory chip prices, which have surged amid AI‑driven demand, are projected to soften in the second half of 2027. Former Samsung Device Solutions head Kye‑hyun Kyung cites expanding Chinese production and a projected global wafer capacity of 6 million units per...
Infineon Expands XHP 2 CoolSiC MOSFET Portfolio with 2300 V Modules for Renewable Energy
Infineon Technologies has added 2300 V CoolSiC™ MOSFET variants to its XHP 2 power‑module family, targeting high‑voltage renewable‑energy converters. The new modules support DC‑link voltages up to 1500 V, feature on‑resistance between 1 mΩ and 2 mΩ, and offer 4 kV or 6 kV isolation. Leveraging silicon‑carbide...
China’s Rapid Chipmaking Expansion Threatens AI Memory Chip Boom, Samsung Adviser Warns
Samsung adviser Kyung Kye‑hyun warned that China’s aggressive chip‑making expansion could blunt the AI‑driven memory super‑cycle by 2028. Chinese firms already hold about 20% of the NAND market and could capture 12‑13% of DRAM after adding 300,000 wafers over three...
LG's UltraGear Is a Native 1,000Hz Full HD Gaming Monitor
LG has launched the UltraGear, a 24.5‑inch Full HD gaming monitor that delivers a native 1,000 Hz refresh rate. Unlike Samsung and Acer’s dual‑mode models, it maintains the 1080p resolution while hitting the ultra‑high refresh speed. The monitor targets competitive FPS players,...
China Market for Nvidia AI Chips to Open 'over Time': Huang
Nvidia CEO Jensen Huang told Bloomberg that China will gradually allow sales of its high‑end H200 AI chip, which is currently licensed but blocked by U.S. export rules. He did not meet Chinese leaders directly, but noted President Trump has...
AMD EPYC Venice Reportedly Combines TSMC N2P and N2 CCD Designs
AMD’s upcoming EPYC “Venice” server line will employ two distinct chiplet (CCD) designs. The mainstream variant will use TSMC’s N2P process, delivering up to 96 cores per socket with 12 Zen 6 cores per CCD. A high‑density version will run on...
Research Bits: May 19
Researchers at the University of Washington unveiled a low‑power, electrically programmable photonic integrated circuit built with standard foundry processes, using phase‑change material to retain settings without power. MIT scientists demonstrated implosion carving to shrink hydrogel‑based optical features from 800 nm to...
Intel Reportedly Urging OEMs Toward Faster Adoption of 18A Consumer CPUs
Intel is reportedly pressing major PC OEMs to accelerate the rollout of consumer processors built on its new 18A manufacturing node. The company aims to shift production capacity from aging Intel 7 and Intel 4/3 chips toward higher‑margin Panther Lake and Wildcat Lake CPUs....
Uviquity Announces World’s First Chip-Scale Deep-UV Laser
Uviquity announced the world’s first chip‑scale deep‑UV laser operating at 229 nm, produced from a single semiconductor chip. The device leverages the company’s aluminum nitride photonic‑integrated‑circuit platform and second‑harmonic generation to deliver collimated, narrow‑line‑width output. Sampling for OEM partners begins in...
Rohm Scales Power Supply Tools for Automotive SoCs
Japanese chipmaker Rohm has begun mass production of a configurable automotive power‑supply platform that pairs its BD968xx‑C series PMICs with the BD96340MFF‑C DrMOS driver. The solution targets system‑on‑chips used in ADAS, driver‑monitoring and in‑vehicle cameras, offering a 2.7‑5.5 V input range...
Titan Army P275MV Plus Gaming Monitor Tested: MiniLED Meets Dual Mode – High Performance, Plenty of Brightness, and a Surprisingly...
Titan Army’s P275MV Plus is a 27‑inch 4K Mini‑LED gaming monitor that blends true HDR performance with a Dual‑Mode that drops to Full HD @ 320 Hz for ultra‑fast competitive play. It offers 1152 local‑dimming zones, 1200 cd/m² peak HDR brightness, 0.5 ms gray‑to‑gray response and 4.2 ms...
ESP32-S3 PowerFeather V2 Board Gains Support for LiFePO4/LFP Batteries
Adafruit’s ESP32‑S3 PowerFeather V2 adds native support for LiFePO4 (LFP) batteries while retaining compatibility with Li‑Ion and LiPo packs. The board incorporates an Analog Devices MAX17260 fuel gauge and a TPS631013 buck‑boost regulator to keep 3.3 V stable across a 5 V‑18 V...
Microchip Ethernet PHY Families Enable Secure, Scalable Connectivity
Microchip Technology introduced the LAN878x and LAN888x families of Single Pair Ethernet (SPE) PHY transceivers, covering 100BASE‑T1, 1000BASE‑T1 and dual‑speed 100/1000BASE‑T1 options. The devices embed hardware‑based MACsec security compliant with IEEE 802.1AE‑2018 and native Time‑Sensitive Networking (TSN) for deterministic, low‑latency communication....
DEEPX Partners with Ultralytics to Define Global Standard for Physical AI
DEEPX and Ultralytics announced a partnership to create a global standard for Physical AI by embedding DEEPX’s DX‑M1 neural processing unit into the Ultralytics YOLO ecosystem. The integration introduces a new “format=deepx” flag that lets developers export YOLO models to...
Variscite to Launch NXP i.MX 95-Based SoM Product Family
Variscite is set to launch a SMARC‑compatible system‑on‑module (SoM) built around NXP’s i.MX 95 processor. The VAR‑SMARC‑MX95 will deliver up to six Cortex‑A55 cores, dedicated real‑time co‑processors, and a 2 TOPS eIQ Neutron NPU for edge AI workloads. Three variants will embed...
Pentagon's Classified AI Push Spurs Surge in Rugged Embedded Computing Demand
The Pentagon announced contracts with SpaceX, OpenAI, Google, NVIDIA, Microsoft, Amazon Web Services, Oracle and others to run frontier AI models inside classified Impact Level 6 and 7 environments. The move is expected to accelerate orders for ruggedized, low‑SWaP embedded...
NVIDIA Ships Vera, Its First AI‑agent CPU, to Anthropic, OpenAI, SpaceXAI and Oracle
NVIDIA handed its first Vera CPUs to Anthropic, OpenAI, SpaceXAI and Oracle Cloud Infrastructure, marking the first production run of a processor designed specifically for AI agents. The 88‑core chip promises 50% faster per‑core performance and 1.2 TB/s memory bandwidth, a...
Google's Willow Chip Achieves Scalable Quantum Error Correction, Cutting Errors to 2.1%
Google announced that its Willow quantum processor achieved scalable error correction, lowering error rates as qubit count grew. The 105‑qubit chip reduced errors from 3% to 2.1% across increasingly larger qubit lattices, a breakthrough that could accelerate practical quantum applications.
Verification IP: Why Design Teams Buy Instead of Build
Verification IP (VIP) offers pre‑built, reusable verification components for standards‑based interfaces and memories, letting design teams avoid the costly effort of building monitors, checkers, and test suites from scratch. As protocols like PCI Express, CXL, DDR, and emerging chiplet interconnects evolve...
Georgia Tech Researchers Unveil Innovative NAND Flash Storage Technology for Deep Space Missions
Georgia Tech researchers have created a ferroelectric NAND flash memory that can survive radiation doses up to one million rads—about thirty times the tolerance of conventional charge‑trapping NAND. The device uses a hafnium‑oxide ferroelectric layer that stores data via polarization...
Uncle Sam's Next Big Supercomputer Might Use Something More Exotic than GPUs
U.S. national labs are testing NextSilicon’s Maverick‑2 dataflow processor as a potential alternative to GPU‑centric supercomputers. Sandia National Laboratory’s 64‑node Spectra testbed, built with Penguin Solutions, recently met all system acceptance criteria, proving the chips can run key HPC workloads....
Cisco Wins Over AI Customers With Merchant Silicon And Optics
Cisco reported Q3 FY2026 revenue of $15.84 billion, a 12% year‑on‑year increase, with operating income climbing 23.7% to $3.96 billion and net income up 35.4% to $3.37 billion. The company raised its AI order outlook to $9 billion for the fiscal year, expecting $4 billion...
'I'm Sorry Dave': NASA Is Working on an AI Chip to Help Next-Generation Spacecraft Think for Themselves — so Clearly...
NASA’s High‑Performance Spaceflight Computing (HPSC) project is unveiling a new radiation‑hardened AI processor that promises up to 100 times the computing power of current spaceflight hardware, with early tests indicating performance as high as 500 times. The multicore chip is engineered to...
AMD Invests $6.5 Million in Marvell, Signaling Chip‑maker Confidence in Data‑center Networking
Advanced Micro Devices disclosed a $6.5 million purchase of 65,516 Marvell shares at $99.05 each, a move analysts say signals confidence in data‑center networking technology. The investment arrives as Nvidia has already poured $2 billion into Marvell and the broader semiconductor rally...
Tata Electronics Inks Partnership with ASML for Gujarat Semiconductor Fab
Tata Electronics and ASML have signed an agreement to establish a semiconductor manufacturing facility in Gujarat's Dholera region, marking a major step in India's push for a self‑sufficient chip supply chain. The partnership, announced during Prime Minister Narendra Modi's Europe...
Ukraine Deploys Home‑Made FP‑1 and RS‑1 Drones in Massive Moscow Strike
Ukraine announced that its forces employed three domestically produced drones—FP‑1 Firepoint, RS‑1 Bars and Bars‑SM Gladiator—in a coordinated assault that saw more than 120 unmanned aircraft over the Moscow region on May 16‑17, 2026. The strike underscores the growing potency...
EnerSys Launches AlphaCell™ 4.0HP+ Battery to Improve Replacement Interval Stability in Outdoor Communications Networks
EnerSys announced the launch of its AlphaCell™ 4.0HP+ battery, a Thin Plate Pure Lead (TPPL) solution built for outdoor and edge communications networks. The battery is engineered to boost electrochemical stability and extend replacement intervals, addressing the growing demand for...
Aramco, Pasqal Deploy Saudi Arabia’s First Quantum Computer and QCaaS Platform
Aramco and French quantum‑computing firm Pasqal inaugurated a 200‑qubit neutral‑atom quantum processor at Aramco’s Dhahran data centre, simultaneously launching the Middle East’s first Quantum Computing as a Service (QCaaS) platform. The partnership aims to accelerate quantum‑enhanced solutions across energy, materials...
Fabricked Attack Undermines AMD SEV‑SNP, Threatening Confidential‑Computing SaaS
Security researchers have demonstrated that a malicious hypervisor can exploit a misconfigured Infinity Fabric to bypass AMD's SEV‑SNP protections. The Fabricked attack compromises the core memory‑access controls of Confidential Virtual Machines, raising urgent concerns for SaaS providers that rely on...
Was My $48K GPU Server Worth It?
In 2024 the author quit a FAANG job and built a $48 K personal GPU server, "grumbl," equipped with six RTX 6000 Ada cards to accelerate independent AI research. By logging utilization and electricity costs, the author found the rig ran at an...
Dell PowerRack Launches at Dell Technologies World 2026 as a ‘Turnkey’ Networking, Storage, and Compute System for AI
Dell Technologies introduced PowerRack at Dell Technologies World 2026, a fully integrated rack‑scale system that combines networking, storage and compute for AI and HPC workloads. The solution promises to go from delivery to live operation in as little as six...
Budget Laptops with Intel Wildcat Lake Launch for Around $450 and Up
Intel introduced the Core Series 3 “Wildcat Lake” processors aimed at budget laptops, now appearing in devices like the Chuwi UniBook priced at $449. The chips share architecture with higher‑end Panther Lake, offering up to six cores (two performance, four efficiency)...
Mouser Electronics and Efficient Power Conversion (EPC) Enter Global Distribution Agreement to Deliver High-Performance Power Solutions
Mouser Electronics has signed a global distribution agreement with Efficient Power Conversion (EPC) to sell EPC's eGaN‑based power devices worldwide. The partnership will make GaN transistors and integrated power stages—spanning 15 V to 350 V—available to engineers designing DC‑DC converters, lidar, motor...
Coordinate Early Risk Identification for Scalable Multi‑Die Assemblies
Multi-die assemblies are facing full system-level challenges, but engineering teams need coordinated and repeatable ways to identify risks early and scale reliably. https://t.co/maXDZJsGsx #chiplets
Fermilab Leads Multi-Lab AI Initiative to Accelerate Design of Chips Used in Extreme Environments
Fermilab is spearheading the AXESS initiative, a DOE‑wide AI effort to accelerate custom micro‑electronics for extreme environments. By integrating large‑language models and surrogate AI, the team aims to shrink chip‑design cycles from months to weeks. Early results show a 500‑fold...
Dell to Support Samsung with Infrastructure for AI-Driven Chipmaking
Dell Technologies announced a partnership with Samsung Electronics to deliver standardized compute, storage, and data‑movement infrastructure for the chipmaker’s AI‑driven semiconductor manufacturing. The solution will underpin Samsung’s digital twin, real‑time analytics, and agentic AI tools across research, design, and production...