ROCKSTAR ELECTRONICS LIMITED’s Post

Update Market Trend Mar 4th -Mar 8th 2023　　　　 #AWD Sales of "customized AI chips" to China were blocked by the U.S. as "too powerful". Launched 16nm FPGA chip portfolio with 30% lower power consumption. #AMD Radeon RX7900 GRE graphics card announced for global release.　　　　 #SAMSUNG Reportedly using NVIDIA's digital twin technology to improve chip yields. Second-generation 3nm process renamed '2nm' for marketing purposes. #Samsung and LG will supply 3.9 trillion won of OLED displays for Apple's iPad Pro.　　　　 #MICRON Announced mass production of #HBM3E high bandwidth memory, which will consume 30% less power than competitors. Will utilize Canon's nanoimprint lithography to reduce DRAM production costs. Introduced UFS in a compact package to help design next-generation smartphones with higher-capacity batteries.　　　　 #NVIDIA Next-generation AI chip B200 will consume 1,000W of power and is expected to come out in 2025. GTX16 series GPUs completely discontinued　　　　 #HYNIX Investing $1 billion to ramp up #HBM packaging. May cooperate with Armor Man to produce HBM in Japan.　　　　 #MEMORY DEVICES TrendForce expects memory prices to continue to rise in #Q1 as restocking demand leads to a slow season. Samsung, SK Hynix and Micron cut capex and NAND production in response to supply/demand imbalance IDC's forecast data shows that the memory chips market will grow by as much as 52.5%.　　　　 #AI CHIPS Qualcomm's Smart Driving Chip Set for Toyota, FAW Red Flag Projects, Mass Production Expected by End of '24 NVIDIA's H100 supply limited, A10080GB inventory improves, pricing may be optimized NVIDIA RTX A4000 to be discontinued by 2027　　　　 #MCU Infineon's latest #MCU TC4XX series started mass production -IAR fully supports Siu Wah's full range of chips, strengthening the industrial control and automotive #MCU ecosystem. MCU chips have opened a new round of price increase recently, but the signal of market demand recovery is not clear.

If they keep curbing components and artificially inflating demand through offer curbs, not only speculation grows, but people will stop buying devices that are already too expensive to many...exactly in Guanghua, I already see people delaying purchases or settling for subpar solutions simply because what they would like has reached unreasonable prices by whatever standard. The key theme of the discussion is only one: "太貴". And this despite the offers, at least on basic configurations that just do not suit many. _______________________________________ In the depths of Guanghua Digital Plaza, Taiwan’s largest electronics market, purchasing agents dashed from one store to another last October, sweeping Nvidia’s RTX 4090 graphics cards on to flatbed trolleys and leaving shelves empty of the premium chips. The buyers unzipped fanny packs filled with wads of cash and paid millions of New Taiwan dollars for the goods, retailers who spoke to Nikkei Asia recalled.“They are buying these to later resell them to areas that faced US restrictions at a much higher price,” one store owner in the market said, adding that graphics cards could be resold for three times the price across the strait. “They can just carry them to Hong Kong or other Chinese cities.” https://lnkd.in/gfb_UdS7

SK hynix TO MASS-PRODUCE GDDR7 DRAM CHIP IN Q3 SK hynix has unveiled its next-generation GDDR7 graphics memory chip, boasting industry-leading performance in speed and power efficiency. The new chip, introduced at GTC 2024 and Computex 2024, will enter mass production in Q3. GDDR7, crucial for high-performance graphics cards and AI workloads, operates at 32 gigabits per second—a 60% improvement over the previous generation—and can reach up to 40 gigabits per second under certain conditions. It processes data at over 1.5 terabytes per second and improves power efficiency by 50%. To tackle heat issues, SK hynix has increased the heat-dissipating substrate layers from four to six and applied epoxy molding compound for better protection. Lee Sang-kwon, head of DRAM product planning, anticipates GDDR7's wide applications in 3D graphics, AI, high-performance computing, and autonomous driving. Competitors Samsung and Micron are also advancing in the GDDR7 market, highlighting the fierce competition in the high-bandwidth memory sector. TrendForce notes that GDDR7, carrying a 20-30% premium over GDDR6, is poised to become the flagship product in the graphics memory market. Thanks again to The Korea Herald for the full article with more background and insights click the source link in the comments below 💡 🙏👇 FRIENDS: Join me at the International Semiconductor Executive Summits global events where we shape the future of our industry. Together, we'll strategize, network, and chart the roadmap for semiconductor innovation. Don't miss this opportunity to connect with fellow leaders and drive our field forward! #Semiconductors #MemoryChips #GDDR7 #SKhynix #AI #HighPerformanceComputing #TechInnovation #GraphicsCards #Computex2024 #GTC2024

[Competition heats up in HBM market as Nvidia endorses Samsung chip] The race for leadership in the burgeoning high bandwidth memory chips market is growing fiercer, as Nvidia revealed its plan to secure orders for the cutting-edge component from Samsung Electronics, the Korean chip giant seeking to outsmart crosstown rival SK hynix. At a media briefing last week, Jensen Huang, the co-founder and CEO of Nvidia, openly endorsed Samsung Electronics and unveiled the company's plan to use the Korean chipmaker's HBM chips in the future. US-based Nvidia is the biggest buyer of advanced HBM chips, which have become a key component for artificial intelligence graphics processing units. "HBM memory is very complicated and the value added is very high. We are spending a lot of money on HBM," Huang said at a media briefing held as part of the company's annual GPU technology conference Tuesday in San Jose, California. Tap below to read full story. #HBM #Memory #Chip #Semiconductor #Nvidia #Samsung #SouthKorea https://lnkd.in/gEKSG5Sz

SK Hynix prepares to mass-produce next-gen graphics chip: South Korean chipmaker SK Hynix revealed plans to start mass-producing its next-generation graphics memory chip this calendar quarter, sporting an operating speed of 32Gb/s. In a statement, it claimed the GDDR7 is the best-performing graphics memory product, with a 60 per cent improvement in speed from the previous generation. The DRAM chip also is 50 per cent more power efficient due to new packaging technology which addresses heat issues stemming from the ultra-fast processing of data. The chipmaker stated it increased the number of heat-dissipating substrates from four to six and applied epoxy moulding compound (EMC) to the packaging material to reduce thermal resistance by 74 per cent, while the size remained unchanged. Sangkwon Lee, head of DRAM product planning and enablement, predicted the GDDR7 would be used in a wider range of applications including high-specification 3D graphics, AI, high-performance computing and autonomous driving. On 26 July, the company announced plans to invest KRW9.4 trillion ($6.8 billion) to build the first fabrication plant in the Yongin Semiconductor Cluster. The post SK Hynix prepares to mass-produce next-gen graphics chip appeared first on Mobile World Live. http://dlvr.it/TBFdpV

Samsung announced their next-generation HBM4 memory, expected for release in 2025. This new memory promises significant improvements over the current HBM3 generation. HBM4 will offer increased capacities, potentially reaching 256 GB compared to the current 192 GB limit. This is achieved by using taller memory stacks (16 layers compared to 12) and potentially even denser memory modules. Additionally, HBM4 is expected to see higher data transfer speeds compared to the current 10 Gbps limit. Another key improvement in HBM4 is the utilization of 3D packaging technology. This allows for more compact designs and potentially reduces production costs compared to the previous generation. This technology is expected to address some of the pricing concerns associated with HBM3. The arrival of HBM4 is expected to bring significant advancements to high-performance computing. We can expect both AMD and NVIDIA to refresh their current GPU lineups to incorporate this new memory standard, enabling even faster and more powerful graphics processing. #samsung #techgiants #chips #memorychips #hbm #hbm4 #semiconductor #semiconducts #semiconductorindustry #semiconductormanufacturing #innovation #technology #technologynews #hardware #computerhardware #nvidia #amd #gpus

How Gpus Drive The Semiconductor Industry Chain Upwards 🎫 🎟 🎆"The amount of HBM (High Bandwidth Memory) we need is enormous. We are currently negotiating with Samsung, SK Hynix, and Micron, and have already received products from these three companies," said Jensen Huang, CEO of NVIDIA. Relying on GPUs, NVIDIA is at its peak, causing the top three global memory chip manufacturers to vie for its favor. 🔮The surge of generative AI has given GPUs a significant platform, driving a wave of updates in the entire semiconductor industry chain. GPUs have become the "lifeline" of the memory market, achieving the remarkable feat of turning losses into profits within a single quarter. However, the impact of GPUs on the chip industry extends beyond this.  🎀The Savior of Memory Chips Turning Losses into Profits 🌠Since 2021, the memory chip industry entered a nearly two-year downturn, leading to continuous declines in the profits of memory chip manufacturers, even resulting in losses. For instance, Samsung's comprehensive operating profit for 2023 was 6.6 trillion Korean won, a decrease of 84.86% year-on-year. SK Hynix had a cumulative operating loss of 7.7303 trillion Korean won and a net loss of 9.1375 trillion Korean won. It was not until the fourth quarter of 2023 that these major memory manufacturers gradually regained profitability. The key turning point was the explosive growth of GPUs. 🌈GPUs require high bandwidth and capacity for memory when processing large amounts of data, especially in high-performance computing, artificial intelligence, and graphics processing. GDDR (a type of memory used for graphics processors and high-performance computing modules) and HBM...... 🛸Learn More:https://lnkd.in/g8pTDHMP #ai #gpu #automotive #semiconductors #integratedcircuits #electroniccomponents #Sensors #storage #ems #oemfactory #icchips #icchip #automotive #aerospace #industrial #medicalcare #energy #military #hardware #5g #consumerelectronics

Advanced Micro Devices (AMD) for the second quarter reported revenue of $5.8 billion up by 9% YoY. The revenue growth was mainly driven by strong data center performance and solid results in the client business.   AMD reported that AI chip sales, included in its Data Center segment, surged 115% year over year to $2.8 billion, driven by a rise in AI GPU shipments.   AMD's primary business involves manufacturing central processors (CPUs) for laptops and servers. Sales for PCs are reported in the company's Client segment, which increased by 49% from the same period last year to $1.5 billion This growth indicates that the PC market is recovering after several years of post-pandemic decline.   However, that was four years ago, and consumers are now looking to replace the PCs they bought at the start of the pandemic. According to IDC, this has led to a 3% year-over-year increase in worldwide PC shipments in the second quarter, marking the second consecutive quarter of growth after eight quarters of declines.   AMD also manufactures chips for gaming consoles and GPUs for 3D graphics. It reported $648 million in revenue in the gaming category, a 59% YoYr decline.   The company’s embedded segment, consisting of products acquired through the Xilinx acquisition in 2022 for industrial customers, reported $861 million in sales, a 41% year-over-year decrease.   AMD CEO Lisa Su stated on a call with analysts that the company experienced “higher than expected” sales of its AI chips, with revenue from its MI300 chips surpassing $1 billion during the quarter.   “As a result, we now expect data center GPU revenue to exceed $4.5 billion in 2024, up from the $4 billion we projected in April,” Su said on the call.   The chipmaker reported a net income of $265 million, or 16 cents per share, compared to $27 million, or 2 cents per share, in the same period last year.   AMD plans to release new AI chips annually: the MI325X in the fourth quarter of this year, the MI350 in 2025, and the MI400 in 2026, as reiterated by the company on its earnings call. Su noted that the MI350 should be “very competitive” with Nvidia’s Blackwell, which was unveiled in March as “the world’s most powerful chip” for AI and recently began sampling to buyers.   Overall, AMD forecasted revenue of $6.7 billion, with a margin of error of plus or minus $300 million. On an adjusted basis, the company forecasted a gross margin of approximately 53.5% for the third quarter, slightly below the estimates of 53.6%.   On July 30th, AMD shares increased by 8% in extended trading.     Visit Us: https://lnkd.in/dbW2rJ75   Capital at Risk | Sophisticated Investors Only    Sources: AMD CNBC William Rhind Manuj Sarpal Paul Marino Benoit Autier Matthew Lamb, CAIA® Gianmarco Roncarolo Evan Ong

MediaTek and NVIDIA have reportedly about a united effort to develop a new processor with the support of artificial intelligence. The new project is expected to be competitive with Apple's M4. The ARM processor for PCs is expected to be finalized by the end of this year. This project can have a considerable influence on the market thanks to the collaboration between NVIDIA and MediaTek. Details of the new project are likely to be announced on June 2 at COMPUTEX 2. The director of NVIDIA will be present at the international exhibition, so some facts and insights on processor development are coming. The collaboration between the two well-known companies will have a considerable impact on the final product, MediaTek will have a significant influence on energy efficiency and AI support, while NVIDIA will influence the power of processors. The processor is going to be produced on the basis of TSMC's 3-nanometer process technology. On the current schedule, development will be finalized in the third quarter of the current year, product testing will begin in the fourth quarter, and full production will be launched at the beginning of 2025. The approximate cost of the upcoming chip will be approximately $300. It also became known about the packaging that will be used for the processor - CoWoS. The new product is expected to significantly increase performance in terms of artificial intelligence and graphics rendering, largely because NVIDIA is a leader in this area. If we analyze the most recent processors from MediaTek, they include more robust cooling systems for laptops and PCs, while smartphone processors do not offer such strong cooling capacity. Almost a year ago, MediaTek and NVIDIA began working together to develop a car chip with artificial intelligence and advanced graphics. #VPS #server #NVIDIA #MediaTek #TSMC

AMD vs. NVIDIA —The Battle of Chip Titans in the Tech Revolution 1/ In the tech world, few rivalries are as intense as AMD vs. NVIDIA . Both companies are pivotal in the semiconductor industry, but their approaches to innovation and market domination differ significantly. 2/ NVIDIA Taiwan ($NVDA): Dominates the GPU market with its advanced graphics and AI capabilities. NVIDIA’s leadership in AI-driven technologies, from data centers to autonomous vehicles, has made it the cornerstone of the AI revolution. 3/ AMD ($AMD): Once the underdog, AMD has risen with its high-performance CPUs and GPUs. The company's Zen architecture has disrupted the CPU market, challenging Intel, while its GPUs are increasingly competitive with NVIDIA’s. 4/ AI & Data Centers: NVIDIA’s GPUs are essential for AI workloads, giving it a massive edge in the booming AI and data center markets. Meanwhile, AMD’s EPYC processors are gaining traction in data centers, offering a cost-effective alternative to Intel’s Xeon chips. 5/ Gaming & Graphics: While NVIDIA leads with its RTX series, AMD’s Radeon GPUs, especially with RDNA 3, are closing the gap, offering great value for gamers. AMD's partnership with gaming consoles like PlayStation and Xbox also provides a stable revenue stream. 6/ Market Position: NVIDIA’s market cap has skyrocketed, reflecting its dominance in AI and deep learning. AMD, however, has shown exceptional growth, particularly in CPU market share, driven by its efficient design and aggressive pricing. 7/ Long-Term Outlook: Both companies are positioned for growth, but NVIDIA’s AI dominance could propel it to new heights, particularly as AI-driven applications continue to expand. AMD’s diverse portfolio, balancing CPUs and GPUs, offers resilience and growth in multiple sectors. 8/ Final Verdict: If you believe in the AI-driven future, NVIDIA might be the bet. For those who prefer balanced growth across multiple sectors, AMD presents a compelling case. Either way, both companies are key players in the tech landscape, shaping the future of computing. #Semiconductors #TechStocks #AI #NVIDIA #AMD #Investing