News Posts matching #RAM

Return to Keyword Browsing

SK hynix Announces 3Q25 Financial Results

SK hynix Inc. announced today that it has recorded 24.4489 trillion won in revenues, 11.3834 trillion won in operating profit (with an operating margin of 47%), and 12.5975 trillion won in net profit (with a net margin of 52%) in the third quarter. The company achieved its highest-ever quarterly performance, driven by the full-scale rise in prices of DRAM and NAND, as well as the increasing shipments of high-performance products for AI servers. In particular, operating profit exceeded 10 trillion won for the first time in the company's history.

As demand across the memory segment has soared due to customers' expanding investments in AI infrastructure, SK hynix once again surpassed the record-high performance of the previous quarter due to increased sales of high value-added products such as 12-high HBM3E and DDR5 for servers. Driven by surging demand for AI servers, shipments of high-capacity DDR5s of 128 GB or more have more than doubled from the previous quarter. In NAND, the portion of AI server eSSD, which commands a price premium, expanded significantly as well. Building on this strong performance, the company's cash and cash equivalents at the end of the third quarter increased by 10.9 trillion won from the previous quarter, reaching 27.9 trillion won. Meanwhile, interest bearing debt stood at 24.1 trillion won, enabling the company to successfully transition to a net cash position of 3.8 trillion won.

Windows 11 Will Start Memory Scans After BSOD to Prevent Future Issues

The land of Windows 11 is finally getting a feature most users will appreciate, with the introduction of the new memory scanning for issues after a blue screen of death (BSOD) happens. "We're introducing a new feature that helps improve system reliability. If your PC experiences a bug check (unexpected restart), you may see a notification when signing in suggesting a quick memory scan," noted Windows Insider Program lead Amanda Langowski. Additionally, the "If you choose to run it, the system will schedule a Windows Memory Diagnostic scan to run during your next reboot (taking 5 minutes or less on average) and then continue to Windows. If a memory issue is found and mitigated, you will see a notification post-reboot."

Microsoft notes that this first wave flags every bug check so they can watch how memory glitches turn into blue screens, and they will refine targeting of these issues in the later updates. At the moment the preview will not run on Arm64 PCs, machines that have Administrator Protection turned on, or any BitLocker setup that boots without Secure Boot enabled. Users that are part of the Windows Insiders Dev and Beta channels will be able to access this feature. Windows 11 Insider Preview Build 26220.6982 (KB5067109) and Windows 11 Insider Preview Build 26120.6982 (KB5067109) are the first in the rollout, so they can beta test the feature before it hits the main stable Windows 11 branch.

Micron Delivers Industry's Highest Capacity SOCAMM2 for Low-Power DRAM in the AI Data Center

In an era of unprecedented AI innovation and growth, the entire data center ecosystem is transforming toward more energy-efficient infrastructure to support sustainable growth. With memory playing an increasingly critical role in AI systems, low-power memory solutions have become central to this transformation. Micron Technology, Inc. today announced customer sampling of 192 GB SOCAMM2 (small outline compression attached memory modules) to enable broader adoption of low-power memory within AI data centers. SOCAMM2 extends the capabilities of Micron's first-to-market LPDRAM SOCAMM, delivering 50% more capacity in the same compact footprint. The added capacity can significantly reduce time to first token (TTFT) by more than 80% in real-time inference workloads.

The 192 GB SOCAMM2 uses Micron's most advanced 1-gamma DRAM process technology to deliver greater than 20% improvement in power efficiency, further enabling power design optimization of large data center clusters. These savings become quite significant in full-rack AI installations, which can include more than 40 terabytes of CPU-attached low-power DRAM main memory. The modular design of SOCAMM2 improves serviceability and lays the groundwork for future capacity expansion.

Crucial Launches DDR5 Pro OC 6400 CL32, Its Fastest Gaming Memory to Date

Crucial, a brand of Micron Technology, Inc., is deepening its commitment to gamers with the debut of Crucial DDR5 Pro Overclocking (OC) 6400 CL32 Gaming DRAM. Engineered for tech enthusiasts and offered in 32 GB kits or 16 GB single modules, this new memory delivers faster speeds, lower latency and a bold new design that looks as good as it performs. This launch cements Crucial's commitment to PC gamers with new 10-nanosecond/CL32-class latency and delivers on feedback related to aesthetics, now with a new elevated heat spreader design.

"Today's AAA games are more immersive, fast paced and demanding than ever. They push systems to their limits with real-time action, intense multitasking and ultra-high frame rates," said Dinesh Bahal, corporate vice president and general manager of Micron's Commercial Products Group. "Gamers need memory that can keep up. Crucial DDR5 Pro OC 6400 CL32 is designed for those who want to win on both performance and style. It helps players load games faster, react instantly and stay ahead of the competition. It's fast, reliable and looks incredible inside your rig."

[Editor's note: Our in-depth Crucial Pro OC DDR5-6400 32 GB CL32 Review is now live]

Compute-In-Memory APU Achieves GPU-Class AI Performance at a Fraction of the Energy Cost

GSI Technology, Inc. (Nasdaq: GSIT), the inventor of the Associative Processing Unit (APU), a paradigm shift in artificial intelligence (AI) and high-performance compute (HPC) processing providing true compute-in-memory technology, announced the publication of a paper led by researchers at Cornell University. Findings confirmed that GSI Technology's APU CIM (Compute-In-Memory) architectures can match GPU-level performance for large-scale AI applications with a dramatic reduction in energy consumption due to high-density and high-bandwidth memory associated with the CIM architecture.

Key findings include:
  • GPU-class performance - The Gemini-I APU delivered comparable throughput to NVIDIA's A6000 GPU on RAG workloads.
  • Massive energy advantage - The APU delivers over 98% lower energy consumption than a GPU over various large corpora datasets, underscoring its efficiency and sustainability.
  • Faster and more efficient than CPUs - The APU's unique design allows it to perform retrieval tasks several times faster than standard CPUs, shortening total processing time by up to 80%.

ASML Reports €7.5 Billion Total Net Sales and €2.1 Billion Net Income in Q3 2025

Today, ASML Holding NV (ASML) has published its 2025 third-quarter results.
  • Q3 total net sales of €7.5 billion, gross margin of 51.6%, net income of €2.1 billion
  • Quarterly net bookings in Q3 of €5.4 billion of which €3.6 billion is EUV
  • ASML expects Q4 2025 total net sales between €9.2 billion and €9.8 billion, and a gross margin between 51% and 53%
  • ASML expects a full-year 2025 total net sales increase of around 15% relative to 2024, with a gross margin of around 52%
  • ASML does not expect 2026 total net sales to be below 2025
CEO statement and outlook
"Our third-quarter total net sales of €7.5 billion and gross margin of 51.6% were in line with our guidance, reflecting a good quarter for ASML. On the technology side, we see litho intensity continue to develop positively as EUV adoption gains momentum, including progress on High NA EUV. In line with our plans to support our customers in the 3D integration space, we shipped ASML's first product serving Advanced Packaging, the TWINSCAN XT:260, an i-line scanner offering up to 4x productivity compared to existing solutions. Finally, our partnership with Mistral AI allows us to embed AI across our entire holistic portfolio, in order to increase the performance and productivity of our systems and the yield of our customers' processes."

Samsung Expected to Power AMD's MI450 Accelerators with HBM4

Samsung is reportedly set to be the main supplier of HBM4 memory to AMD's upcoming MI450 chips according to TrendForce citing SeDaily. The rumors (news) came just days after AMD announced a multi-year deal with OpenAI paving the way for a 6-Gigawatt partnership (and the option for OpenAI to acquire up to 10% of AMD). Collaboration between Samsung and AMD is nothing new as current generation chips such as MI350X and MI355X feature 12-high HBM3e memory from Samsung and Micron.

OpenAI made initial agreements with Samsung and SK hynix to obtain memory for its huge Stargate data center project last week, before joining forces with AMD. Reuters reports that OpenAI's needs might soar to 900,000 DRAM wafers - a number that could make up almost 40% of the world's total DRAM production. Maeil Business Newspaper states that Samsung plans to provide not just HBM but also graphics DRAM (GDDR) high-capacity SSDs, and LPDDR5X-PIM - low-power DRAM with in-memory computing - to use in AI applications for OpenAI's Stargate project. The industry seems to polarize even more around NVIDIA and AMD since Intel just teamed up with Team Green. To counter NVIDIA's proprietary NVLink, AMD is pushing for the Ultra Accelerator Link (UALink) open-standard alternative. This initiative managed to gather quite a sizable momentum, gathering partners such as Apple, Google, Microsoft, Synopsis, Meta, and Intel. The latter's presence in the UALink consortium remains to be seen, due to its latest partnership with NVIDIA.

Thermal Grizzly Introduces New Minus Pad extreme 2 and Minus Pad High Compression Thermal Pads

Thermal Grizzly introduces the successor to the minus pad extreme: Minus Pad extreme 2. Moving forward, Minus Pad extreme 2 will lead the TG thermal pad portfolio in terms of thermal conductivity. The most significant difference compared to the minus pad extreme is a new material composition. This results in an optimized price-performance ratio while maintaining the outstanding thermal conductivity at a level comparable to that of the minus pad extreme.

The Minus Pad High Compression thermal pads feature exceptionally high compressibility. This makes them particularly versatile in applications where a single pad needs to bridge varying height differences simultaneously. This is especially useful when modifying graphics cards with a GPU water block or when replacing thermal pads on GPU coolers, as a Minus Pad High Compression (HC Pad) can be used across different components such as VRAM, voltage regulators, and other contact surfaces.

Samsung and OpenAI Announce Strategic Partnership To Accelerate Advancements in Global AI Infrastructure

OpenAI, Samsung Electronics, Samsung SDS, Samsung C&T and Samsung Heavy Industries today announced a letter of intent (LOI) for their strategic partnership to accelerate advancements in global AI data center infrastructure and develop future technologies together in relevant fields. This expansive collaboration will bring together the collective strengths and leadership of Samsung companies across semiconductors, data centers, shipbuilding, cloud services and maritime technologies.

The signing ceremony was held at Samsung's corporate headquarters in Seoul, Korea, attended by Young Hyun Jun, Vice Chairman & CEO of Samsung Electronics; Sung-an Choi, Vice Chairman & CEO of Samsung Heavy Industries; Sechul Oh, President & CEO of Samsung C&T; and Junehee Lee, President & CEO of Samsung SDS.

Tight Upstream Supply and Restocking Drive 2024 DRAM Module Revenue Growth of 7%

TrendForce reports that following the completion of inventory digestion in the downstream consumer market concluded at the end of Q4 2023, DRAM suppliers shifted focus towards HBM and server DDR5 products, leading to tighter supply for other DRAM types. This increase in demand pushed overall DRAM prices higher, encouraging module manufacturers to replenish stocks and boost procurement. Consequently, the global DRAM module market achieved USD 13.3 billion in revenue in 2024, a 7% YoY increase that reversed the 28% decline experienced in 2023.

In the latter half of 2024, demand momentum declined due to rising module prices. Module manufacturers struggled to transfer these high chip costs to distribution channels, limiting sales. To stay competitive, some aimed to enhance costs structure by increasing their purchase of cheaper server reball chips.

Raspberry Pi Unveils the new Raspberry Pi 500+ for $200

Meet the newest member of our family of all-in-one PCs: Raspberry Pi 500+. It's a complete desktop computer, a love letter to the machines of our childhoods, and our most polished product yet. Raspberry Pi 500+ boasts a high-quality mechanical keyboard with removable keycaps and individually addressable RGB LEDs, an internal M.2 socket pre-fitted with a 256 GB Raspberry Pi SSD, and 16 GB of RAM. Read on for the full story, or skip straight to the Raspberry Pi 500+ product page to order yours.

Raspberry Pi 500+ is built on the Raspberry Pi 5 platform, featuring a 2.4 GHz quad-core Arm Cortex-A76 CPU, dual 4K display output, dual-band Wi-Fi and much more, and is priced at $200. It is also available in a $220 Desktop Kit, which adds:
  • A Raspberry Pi Mouse
  • A Raspberry Pi 27 W USB-C Power Supply
  • A 2 m micro HDMI to HDMI cable
  • A copy of the Raspberry Pi Beginner's Guide, 5th Edition

DRAM Prices to Continue Rising in 4Q25, Server Demand Surges Ahead While Legacy Process Products See Steeper Increases

TrendForce's latest investigations reveal that the three major DRAM suppliers continue to allocate advanced process capacity primarily to high-end server DRAM and HBM, crowding out capacity for PC, mobile, and consumer applications. Combined with diverging demand across end products, this dynamic is keeping legacy-process DRAM price hikes pronounced, while newer-generation products are seeing more modest gains. Overall, conventional DRAM prices are expected to rise 8-13% QoQ in 4Q25, and when HBM is included, the increase could widen to 13-18%.

PC DRAM prices edge up slightly, U.S. CSPs consider early server DRAM procurement
Shipments of finished PCs are projected to decline with PC promotions and restocking momentum slowing in 4Q25, leading major OEMs to cut DRAM bit procurement. Meanwhile, as the three suppliers aggressively shift capacity toward server DDR5, the supply of both PC DDR5 and DDR4 remains constrained. As a result, PC DRAM prices are forecast to continue edging upward through the quarter.

GMKtec Launches NucBox M5 Ultra Mini PC with AMD Ryzen 7 7730U

GMKtec today announced the global availability of its latest high-performance Mini PC, the NucBox M5 Ultra, on the Global website and European website. Designed for professionals, creators, and everyday users seeking performance in a compact form, the M5 Ultra delivers cutting-edge specifications with flexible configurations and worldwide power compatibility (Japan, US, EU, UK, AU).

The NucBox M5 Ultra is available in multiple configurations - Barebone, 16 GB RAM + 512 GB SSD, and 32 GB RAM + 1 TB SSD - allowing users to choose the ideal setup for their specific performance needs.

China's AI Chip Ambitions Limited by HBM Memory Supply, Notes Report

According to SemiAnalysis, Chinese AI semiconductor buildout faces an unexpected bottleneck that's more critical than manufacturing constraints: high-bandwidth memory shortages. While domestic foundries like SMIC can produce sufficient AI processors, the lack of HBM supply chains has emerged as the primary limitation throttling China's semiconductor ambitions. Huawei's Ascend 910C illustrates this mismatch. The company has foundry capacity to manufacture 805,000 units annually through TSMC, which is a result of stockpiling before the cutoff, as well as SMIC. However, HBM shortages prevent realizing this production potential. Chinese firms accumulated roughly 11.4 million HBM stacks from Samsung before export controls tightened, but these stockpiles won't sustain long-term AI scaling. The memory crunch effectively neutralizes Chinese manufacturing capabilities, giving Western competitors like NVIDIA and AMD a sustained advantage despite Beijing's substantial semiconductor investments.

The path toward HBM independence runs through domestic memory manufacturers like CXMT and the recent collaboration with YMTC. Converting standard DRAM production to HBM manufacturing requires specialized equipment, which is predominantly supplied by Western suppliers, creating another dependency layer. China could achieve competitive HBM3E production by 2026 if current investment trajectories continue and equipment restrictions remain stable. However, this timeline assumes successful technology transfer and uninterrupted access to critical manufacturing tools. If export controls expand to encompass additional HBM-related equipment, the development window could become considerably longer. YMTC is reportedly ready to enter DRAM production and could begin purchasing equipment by late 2025, utilizing its secret weapon, Xtacking technology, one of the most advanced hybrid bonding techniques available. This matters because HBM requires stacking at least 16 memory chips with incredible precision, something YMTC has been doing for years with their NAND products.

Samsung's GDDR6 Modules Run 10°C Cooler Than SK Hynix, Claims GPU AIB

XFX has officially published various performance figures claiming that its custom GPUs run better with Samsung GDDR6 memory, rather than SK Hynix modules. Usually, add-in-board (AIB) partners receive the GPU die sold as a kit with memory. However, AMD, NVIDIA, and Intel typically source their memory modules from various partners, including Samsung, SK Hynix, and Micron. GPU AIBs can sometimes prioritize a single memory maker, and this time it looks like Samsung GDDR6 memory modules are turning out to be a much better design for the XFX GPUs. "In terms of hardware, memory power consumption and heat dissipation also directly impact overall stability. Previously, the XFX 9060 XT used Hynix memory, but this time, the XFX 9060XT V3 has switched to Samsung memory. Not only is the GPU cooler, but its DDR temperature is 10°C lower than the Hynix version" says XFX on BiliBili.

The performance comparisons show that Samsung's GDDR6 memory runs at 77°C, while the SK Hynix modules run at 10°C hotter, 87°C. Switching to Samsung-made modules also brings down the total board consumption to 183 W, down from 207 W with SK Hynix. This is 24 W of heat, which the cooler system picks up, resulting in a lower fan speed. In contrast, the Samsung version spun the fans at 1461 RPM, while the SK Hynix version spun at 1814 RPM. This means that the overall GPU system is being more stable, generally more effective, prevents VRAM downclocks, and has higher stability. Interestingly, the choice of Samsung memory increases the GPU die temperature by 3°C, indicating that the memory controller on the die is likely operating at a higher intensity, or the lower fan speed resulted in less cooling. While we await more details and independent testing, it is interesting to see just how much a memory vendor choice can impact the overall GPU performance. There is little information on XFX 9060 XT V3 design, but it is already available in China.

YMTC and CXMT to Jointly Accelerate China's HBM Production

Two of the biggest names in the Chinese memory/storage industry, YMTC and CXMT, are reportedly teaming up to accelerate the domestic high-bandwidth memory development. YMTC, best known for making NAND flash storage, is apparently ready to jump into DRAM production and could start buying equipment by late 2025. What makes this interesting is YMTC's secret weapon: their Xtacking technology, which TechInsights calls one of the best hybrid bonding techniques out there. This matters because HBM requires stacking at least 16 memory chips with incredible precision, something YMTC has been doing for years with their NAND products. Meanwhile, CXMT has been ramping up fast, pushing out 70% more DRAM wafers than last year and aiming to get HBM3 certified soon, with HBM3E coming within two years.

CXMT already produces HBM2 and has the DRAM manufacturing infrastructure, while YMTC's bonding expertise could help solve the technical headaches that come with stacking memory chips higher and higher without them overheating. Additionally, companies like Wuhan Xinxin and Tongfu Microelectronics are developing the packaging and assembly capabilities necessary to transform these chips into finished products. CXMT expects to produce 2.73 million wafers in 2025, up from 1.62 million this year, and they could soon match the volume of Micron's shipments. However, all we need to watch now is the feasibility of this HBM development, as Chinese companies require cutting-edge DRAM for their accelerators, like Huawei's Ascend. If global demand persists, we may even see some Western customers use HBM originated in China.

32 GB of RAM Could Become the New Standard for Gamers, According to Steam Survey

The Steam Hardware and Software Survey has delivered yet another update, showing us where the general PC gamer is headed. This time, the sector shows an interesting growth is system memory, particularly among users with 32 GB of RAM. While 16 GB of RAM is still the most common configuration, accounting for 41.88% of all surveyed systems, the next most common configuration is actually 32 GB, at 36.46%, with a growth of +1.31% in August compared to July. At this rate, if the growth continues, the 32 GB systems could be on a path to becoming a standard in gamers' choice in just a couple of months, meaning that more users than ever could be on 32 GB RAM systems. DDR5 memory is becoming cheaper, and OEMs are offering more memory configurations, all of which is making gamers opt for higher capacity, whether it's a prebuilt, DIY, or laptop system.

A few familiar forces drive that shift. Current generation platforms from both Intel and AMD utilize dual channel DDR5 configurations, so new builds moving forwards typically start at 32 GB from a pair of 16 GB modules, as densities were doubled from DDR4 to DDR5. Single channel memory from a single installed DIMM would typically be the only way to get just 16 GB of RAM in current generation desktops, since 8 GB modules are quite rare. Modern AAA games increasingly recommend 32 GB for the smoothest experience at high or ultra settings, and anyone who streams, records, or runs background tools alongside a game benefits from the extra headroom. Local AI is part of the equation for some users. Many hobbyist AI tasks rely on GPU VRAM, but certain CPU-based or very large-model setups can also increase system RAM requirements. High-end desktop configurations are a niche example, but show just how much memory some workloads can demand. Students, content creators, and professionals who use rendering, CAD, or EDA alongside gaming likewise favor bigger capacities. Finally, many gamers buy extra RAM to future-proof their systems so they can upgrade a GPU later without worrying about memory.

Getac's New V120 Fully Rugged Laptop Marks a Major Leap in AI-Ready Mobile Computing for Diverse Field Professionals

Getac Technology Corporation (Getac), a leading provider of rugged computing and mobile video solutions and a manufacturer with advanced in-house capabilities, today announced the launch of the V120, a new fully rugged laptop designed to deliver a premium computing experience in challenging work environments. The new V120 is aimed at professionals in industries including defense, automotive, public safety, and utilities, who need devices they can rely on for a diverse range of operational tasks, from mobile command and control to advanced data collection and analysis in the field.

Evolved from Experience, Reinvented Inside and Out
Shaped by years of proven deployment across key industries, the V120 builds on Getac's top-selling V110, incorporating real-world feedback into a refined convertible-hinge design that allows the device to function like a tablet, meeting the evolving needs of frontline professionals.

2Q25 DRAM Revenue Jumps 17.1%, SK hynix Expands Market Share as Taiwanese Vendors Deliver Strong Growth

TrendForce's latest investigations demonstrate that global DRAM industry revenue reached US$31.63 billion in the second quarter of 2025, up 17.1% QoQ. This growth was fueled by rising contract prices for conventional DRAM, robust shipments growth, and expanding HBM volumes. Inventory digestion at DRAM suppliers accelerated with stronger procurement momentum from PC OEMs, smartphone makers, and CSPs, driving contract prices for most products back into positive territory.

SK hynix posted bit shipments above its targets in 2Q25. However, the higher share of lower-priced DDR4 shipments limited ASP growth. Even so, revenue surged 25.8% QoQ to $12.23 billion, boosting market share to 38.7% and securing the top spot once again.

Colorful Reveals DDR5-6400 RAM Overclocked to 9600 MT/s

Colorful claims that it can run a pair of its 16 GB iGame Shadow II DDR5-6400 memory modules in dual-channel at up to 9600 MT/s on an AMD Ryzen 7 9800X3D with a pretty stark overclock. This particular result was achieved with slightly stretched out timings, at CL30 vs the stock 28, and by running the UCLK and MCLK with the usual 1:2 ratio. Complete stability was achieved by dropping the overclock to 9200 MT/s with CL32, suggesting there was some instability at the bleeding edge. Read more about AMD Zen 5 and DDR5 in our article "DDR5 Memory Performance Scaling with AMD Zen 5"

Colorful's testing was validated in RunMemTestPro, and Colorful ran a third test, this time trying to chase lower latencies by tightening up the timings, reducing the memory clock, and running the UCLK and MCLK in 1:1 mode, usually reserved for DDR5 RAM running at 6000 MT/s or less. For this test, Colorful swapped the CPU for a more powerful—and likely better-binned—AMD Ryzen 9 9950X3D, and it managed to achieve 7200 MHz with timings of 24-37-32-28 and 100% completion of the memory test, suggesting decent stability, even with the increased UCLK frequency. The overclocking team also managed to stabilize an overclock with 8900 MT/s and 34-47-47-47 timings. It should be noted that Colorful hasn't actually provided any benchmark testing results to show any performance improvements over the stock 6400 MT/s for this particular RAM kit, nor has it provided any of the voltages or temperature consequences that came about as a result of these overclocks, so it's unclear what the long-term viability or implications of these overclocks would be.

DDR4 and LPDDR4 Supply Tightens Sharply in 2H25 as Structural Shortage Drives Strong Price Surge

TrendForce's latest findings reveal the DDR4 market is set to remain in a persistent state of undersupply and strong price growth through 2H25. Rigid server orders are crowding out supply for the PC and consumer markets, forcing PC OEMs to accelerate DDR5 adoption.

Manufacturers of consumer electronics, meanwhile, are struggling with high prices and material shortages. The tight DRAM supply-demand balance has also pushed up mobile DRAM contract prices, with 3Q25 LPPR4X posting the largest single-quarter increase in nearly a decade.

Corsair Reports Strong Q2 2025 with Double-Digit Year-Over-Year Growth in Revenue and Gross Profit

Corsair Gaming, Inc. (Nasdaq: CRSR) ("Corsair" or the "Company"), a leading global provider and innovator of high-performance products for gamers, streamers, content-creators, gaming PC builders and SIM driving enthusiasts, today announced its financial results for the second quarter ended June 30, 2025. Corsair delivered another strong quarter, achieving double-digit growth in revenue and gross profit, expanded margins, and broad-based demand strength across its entire product portfolio. Results exceeded expectations, underscoring the Company's strengthened leadership position in the gaming and content creator ecosystems.

Q2 2025 Financial Highlights (compared to Q2 2024)
  • Revenue increased 23% to $320 million, above consensus.
  • Gross profit increased 36% to $85.9 million, reflecting improved product mix, favorable channel performance, and increased operational efficiency.
  • Adjusted EBITDA improved to $8.1 million from a loss of $1.2, also above consensus
  • Net Income (loss) per share attributable to common stockholders improved to $(0.16) / $0.01 per diluted share on a GAAP / Non-GAAP basis, compared to $(0.28) / $(0.07), respectively.

NEO Semiconductor Introduces World's First Extreme High Bandwidth Memory (X-HBM) Architecture for AI Chips

NEO Semiconductor, a leading developer of breakthrough memory technologies, today introduced the world's first Extreme High Bandwidth Memory (X-HBM) architecture for AI chips. Built to meet the growing demands of generative AI and high-performance computing, X-HBM delivers unmatched performance with a 32K-bit data bus and potentially 512 Gbit per die, dramatically surpassing the limitations of traditional HBM with 16X greater bandwidth or 10X higher density.

"X-HBM is not an incremental upgrade, it's a fundamental breakthrough," said Andy Hsu, Founder & CEO of NEO Semiconductor. "With 16X the bandwidth or 10X the density of current memory technologies, X-HBM gives AI chipmakers a clear path to deliver next-generation performance years ahead of the existing roadmap. It's a game-changer for accelerating AI infrastructure, reducing energy consumption, and scaling AI capabilities across industries."

ADATA Trusta Highlights SSD Power Efficiency for AI Servers at OCP APAC 2025

TRUSTA, the enterprise storage brand of memory leader ADATA, announced its first-ever participation in the Open Compute Project APAC Summit, the annual technical conference for the open compute industry in Asia-Pacific region, debuts in Taiwan this year, gathering industry leaders to focus on innovations in data centers and AI infrastructure. In collaboration with FADU, a leading enterprise SSD controller brand, TRUSTA is showcasing their latest technological achievements, presenting a key solution for the sustainable development of data centers in the AI era.

TRUSTA Emphasizes SSD Power Efficiency Amid AI Boom and Corporate Sustainability Challenges
With the explosion of AI applications, data centers are facing unprecedented storage pressure. Since 2012, the computing power required for large language models has grown 4.7 times annually. Industry forecasts indicate that by 2030, the data volume for AI inference will surpass that of training data by 3.8 times, with energy consumption demand reaching 1.65 times that of 2023. At the same time, generative AI is accelerating the widespread adoption of synthetic data to surpass real-world data. The resulting data explosion—combined with the 24/7 operation of modern data centers—is intensifying the conflict between technological innovation, energy demand, and carbon emissions.

Enfabrica Unveils Industry's First Ethernet-Based AI Memory Fabric System

Enfabrica Corporation, an industry leader in high-performance networking silicon for artificial intelligence (AI) and accelerated computing, today announced the availability of its Elastic Memory Fabric System "EMFASYS", a transformative hardware and software solution designed to dramatically improve the compute efficiencies in large-scale, distributed, memory-bound AI inference workloads. EMFASYS is the first commercially available system that integrates high-performance Remote Direct Memory Access (RDMA) Ethernet networking with an abundance of parallel ComputeExpressLink (CXL) based DDR5 memory channels. The solution provides AI compute racks with fully elastic memory bandwidth and memory capacity, in a standalone appliance reachable by any GPU server at low, bounded latency over existing network ports.

Generative, agentic, and reasoning-driven AI workloads are growing exponentially—in many cases requiring 10 to 100 times more compute per query than previous Large Language Model (LLM) deployments and accounting for billions of batched inference calls per day across AI clouds. The EMFASYS solution addresses the critical need for AI clouds to extract the highest possible utilization of GPU and High-Bandwidth-Memory (HBM) resources in the compute rack while scaling to greater user/agent count, accumulated context, and token volumes. It achieves this outcome by dynamically offloading HBM to commodity DRAM using a caching hierarchy, load-balancing token generation across AI servers, and reducing stranding of expensive GPU cores. When deployed at scale with Enfabrica's EMFASYS remote memory software stack, the solution enables up to 50% lower cost per token per user, allowing foundational LLM providers to deliver significant savings in a price/performance tiered model.
Return to Keyword Browsing
Nov 16th, 2026 04:24 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts