According to South Korean media Newdaily, Samsung is raising prices for its LPDDR4X, LPDDR5, and LPDDR5X memory products by 30%, and NAND flash memory products like eMMC and UFC by 5-10%. The tight supply is primarily due to production cuts for older products and growing demand from large cloud service providers.

This round of price increases reflects a structural shift in the memory industry. Manufacturers are shifting their focus to emerging markets like AI PCs and next-generation smartphones, leading to a contraction in the supply of traditional products. Morgan Stanley predicts that as competition in the HBM market intensifies, traditional DRAM and NAND products are expected to see more sustainable growth by 2026.

Samsung isn’t alone in this action. Micron has notified customers of a 20-30% price increase and has suspended new orders. SanDisk has also announced a 10% price increase for its NAND flash memory products. This demonstrates the pressures facing the entire memory industry from a supply-demand imbalance.

The fundamental reason for the tight supply is a shift in industry focus. With the adoption of the LPDDR5/X standard in AI PCs and next-generation smartphones, manufacturers have reduced production of older products, but production capacity for the new standard has yet to keep pace with growing demand. DDR4 memory prices have skyrocketed by 50%, making DDR5 a more cost-effective PC solution.

Rising memory prices will directly impact procurement costs for consumer electronics and businesses. As one of the world’s largest memory manufacturers, Samsung’s pricing strategies often lead the industry, leading to market concerns that supply constraints could persist until 2025.

The surge in demand for high-bandwidth memory (HBM) has become a key factor driving up prices. Major DRAM manufacturers have shifted their focus to AI, prioritizing the supply of their latest products for AI accelerators from companies like NVIDIA and AMD. This shift in priorities has led to a tighter supply of consumer-grade DRAM.

Samsung currently holds a 32.7% share of the DRAM market and a 32.9% share of the NAND market. The company is working to secure support from Nvidia to promote its HBM products and is accelerating the development of LPDDR6 DRAM, with the first designs expected to launch later this year.

Morgan Stanley previously stated in a research report that HBM’s “premium myth” faces challenges, with intense competition and pricing pressure expected in 2026. Amidst expectations of Federal Reserve rate cuts and an improving macroeconomic environment, market funds are shifting from AI-driven HBM to traditional storage, with Morgan Stanley particularly bullish on the investment value of traditional DRAM.

The tightening supply and demand structure for DRAM and NAND flash memory appears to be reinforcing the need for price increases. Citigroup predicts that DRAM and NAND flash memory supply will be short by 1.8% and 4%, respectively, next year. Morgan Stanley also predicts that NAND supply will be short by as much as 8%.

The post Samsung reportedly hiked memory and flash memory prices by as much as 30% appeared first on INCUS IT - IT Hardware Distributor.

Scott Davis, senior vice president and chief sales and marketing officer of Western Digital, said in the document that the price adjustment was based on the unprecedented market demand for the entire line of mechanical hard drives , with the aim of supporting business growth, maintaining quality, and continuing to provide customers with lower TCO.

In addition, Western Digital will increase the proportion of sea transportation in product logistics , which may extend the product transportation cycle by 6 to 10 weeks. Users need to reserve sufficient time when placing orders.

The post Western Digital announced that it will gradually increase the prices of all mechanical hard drive products starting today. appeared first on INCUS IT - IT Hardware Distributor.

Samsung has decided to extend the production period of DDR4 by one year. Photo/Associated Press

DRAM giant Samsung Electronics was scheduled to discontinue DDR4 production by the end of this year, but due to soaring DDR4 prices due to strong demand, Samsung decided to extend the production period of DDR4 by one year until the end of 2026.

Samsung previously informed customers that it would phase out production of 1z nanometer process DDR4 DRAM by the end of 2025, but Korean media “The Elec” quoted insiders as reporting that Samsung recently abandoned the plan and decided to continue producing DDR4 until December 2026. Samsung will notify customers of the changes in the plan in the near future.

One of the reasons why Samsung has extended the life of DDR4 products is that the company lags behind competitors SK hynix and Micron in high-bandwidth memory chips (HBM) in emerging fields, limiting HBM’s contribution to performance.

Samsung has decided to make good use of its existing production capacity and continue to leverage the value of DDR4 production lines. The 1z production line has been depreciated, so the cost is low, and continuing to produce DDR4 can greatly increase the profitability of the chip business.

The jump in DDR4 prices is another consideration for Samsung. As companies such as SK hynix and Micron gradually withdraw DDR4 and shift production capacity to HBM, which is widely used in AI applications, the price of DDR4 has skyrocketed due to reduced market supply. In addition, some device manufacturers are stockpiling DDR4 based on US tariff policies, which also supports prices.

According to the latest report from market research firm DRAMeXchange1, the average price of 8GB DDR4 for PCs reached $3.9 in July, a 50% increase from June, setting a record for four consecutive months of increase. From April to June, the price increases were above 20%, 22.22%, 27.27% and 23.81% respectively.

Analysts at data provider TrendForce pointed out that the world’s three major DRAM suppliers, Samsung, Hynix and Micron, are facing supply pressure because the industry prioritizes server DRAM demand rather than PC DRAM.

Entering the third quarter, the price of DDR4 is already about 4% higher than DDR5. According to TrendForce data, the price of an 8GB DDR4 module combination reached about $26.5 in July, up from $25.5 for DDR5 modules.

Samsung’s decision to extend the production period of DDR4 is expected to alleviate market shortages. Sources said that while DDR4 prices are unlikely to fall, price increases are expected to slow down. Furthermore, the continued production of DDR4 also indirectly proves that Samsung’s progress in expanding HBM production is not as smooth as expected.

The post The price soared Samsung continued to produce DDR4 until the end of 2026 appeared first on INCUS IT - IT Hardware Distributor.

Kioxia has introduced a solid-state drive with an unprecedented storage capacity of 245.76TB, currently the highest-capacity storage device available. The company positions its LC9 drive for applications that benefit from maximum storage density, such as AI training and hyperscale workloads. Remarkably, the company has managed to cram 8TB of storage capacity into a single NAND package —a record that enables the construction of such a hefty SSD using multiple ultra-high-capacity packages. To give some perspective on the sheer amount of storage, assuming a heavily compressed 20GB size for a 4K movie, the Kioxia drive could store an astounding 12,500 movies per device.

The Kioxia LC9 is based on an unknown proprietary NVMe 2.0-compliant controller with a PCIe 5.0 x4 interface (with dual-port capability) as well as Kioxia’s BiCS8 2Tb 3D QLC NAND memory devices. To enable a 245.76TB capacity in an E3.L form-factor, Kioxia had to package 32 memory devices into a single chip (8TB of storage space) and then attach those chips to the controller. This packaging sets a new record for density and uses the company’s proprietary methods for wafer processing and assembly, according to Kioxia.

Performance-wise, the drive delivers sequential read speeds of up to 12 GB/s and write speeds of up to 3 GB/s. This isn’t the highest performance for an enterprise PCIe 5.0 SSD, but there’s an obvious tradeoff between capacity and performance. At the end of the day, ensuring signal integrity in a 32-Hi 3D NAND package is a big deal, so it is not surprising that it is achieved by sacrificing interface speed.

For random operations, the SSD achieves up to 1.3 million IOPS for reads and 50,000 IOPS for writes, which is again not particularly impressive, but the LC9 is clearly designed for capacity rather than performance.

As for endurance, the Kioxia L9 drives are rated at 0.3 drive writes per day (DWPD), which again reflects a balance between capacity and longevity for data center usage patterns.

data, including die-level recovery, parity-based error management, and safeguards against sudden power loss. It also includes support for Flexible Data Placement, which reduces internal write amplification and extends the usable lifespan, especially in database-driven applications.

Security options include encryption and firmware signing compliant with CNSA 2.0 standards. The drive supports multiple encryption modes, including AES-256 encryption and algorithms prepared for post-quantum cryptography.

The post Kioxia unveils 245TB SSD, the world’s highest capacity storage device – can store 12,500 4K movies appeared first on INCUS IT - IT Hardware Distributor.

Electronic Times reported that South Korean memory chip makers such as SK Hynix and Samsung Electronics have suspended plans to phase out DDR4 chips and may continue to supply until spot prices lag significantly behind DDR5. At the same time, the two companies will remain cautious in supply to ensure that there is no excess inventory.

The post Samsung and SK Hynix postpone DDR4 memory phase-out due to unexpected demand appeared first on INCUS IT - IT Hardware Distributor.

Intel CEO Chen Liwu admitted to employees that the company is not even among the top ten semiconductor companies in the world. (Data photo, provided by Intel)

Intel used to be the global chip hegemon, but in recent years, in addition to the slump of the foundry business, it has missed the wave of AI, resulting in a wider gap with its rivals. The foreign media “The Oregonian” reported that Intel CEO Chen Liwu admitted to employees that the company is not even ranked among the top ten semiconductor companies in the world, especially in AI training, “Nvidia is too strong”.

According to the report, Chen Liwu said in an internal live broadcast for employees around the world, “20 or 30 years ago, we were indeed leaders, but now the world has changed. We’re not even among the top 10 semiconductor companies in the world.”

Chen Liwu bluntly said that customers gave Intel a failing evaluation, and the company is far behind the industry leader Nvidia in the development of AI training technology, and it is difficult to catch up. He described Intel’s transformation as a marathon, and that the layoffs launched this week are designed to make Intel more agile and swift like Nvidia, Broadcom and AMD, “and we have to be humble.”

The Oregonian confirmed Chen’s statement to Intel, and the spokesperson clarified that Chen’s “top 10” statement refers to market capitalization, not technology leadership. However, the report pointed out that during the 20-minute conversation, Chen himself never mentioned market capitalization, and he talked about the topic of “Big 10 companies” as a response to a question about Intel’s corporate culture.

The report mentions that in fact, the status of Intel’s chip industry has indeed declined seriously, and the current market value is about $100 billion, only half of what it was 18 months ago; In contrast, Nvidia once broke through the $4 trillion market capitalization mark this week, becoming the first company in history to reach this scale. Chen Liwu launched a major layoff plan, saying that “we still have a lot of work to do”, Intel’s market share in the data center market has been seriously lost, and although the PC business has “improved slightly”, the company must strengthen its architecture to meet the needs of high-end computing.

According to the report, the rise of AI technology has worsened Intel’s predicament. At present, AI training relies on NVIDIA’s GPUs, and Intel itself does not have high-end GPU products that can compete with it, and it has been almost marginalized in this wave of AI. Nvidia’s market position is too strong.”

Chen Liwu revealed that Intel will instead focus on “edge AI”, that is, let AI computing power be directly executed on personal computers and terminal devices, rather than concentrated in large data centers, and also hopes to continue to explore the emerging field of “agentic AI”, that is, AI can operate independently without continuous instructions from humans, “This is an emerging field with great potential, and we must grasp it.”

The post Tell the truth? Chen Liwu said frankly that Intel “can’t even rank in the top 10 companies”: Nvidia is really too strong appeared first on INCUS IT - IT Hardware Distributor.

According to Digitimes , Micron’s executive vice president and chief business officer Sumit Sadana confirmed in an interview that Micron has sent a notice to customers, informing them that its DDR4 and LPDDR4 will enter the EOL (End of Life) stage, and the supply is expected to gradually decrease in the next 6 to 9 months.

Sumit Sadana admitted that the market demand for DDR4 is still high, and the gradual discontinuation of production will cause a shortage of DDR4 and LPDDR4 supply, and the market price will even be higher than some DDR5 products in some cases. In order to solve the problem of supply imbalance, Micron will give priority to supplying long-term partners in the automotive, industrial and networking fields, where product supply stability and reliability are crucial. On the other hand, it is expected that as AI functions become more and more common, the demand for memory in PCs and mobile devices will increase moderately.

Although major memory manufacturers have begun to shift to DDR5 and LPDDR5 on a large scale, the market does not seem to be ready yet. Some smaller suppliers may get more opportunities in DDR4 and LPDDR4, but they may not have the scale and experience of major memory manufacturers, resulting in lower product quality and reliability.

The post Micron has notified customers that DDR4 has entered the EOL stageSupply is expected to gradually decrease in the next 6 to 9 months appeared first on INCUS IT - IT Hardware Distributor.

Today, DDR4 prices are higher than those of higher-specification DDR5, showing a “price inversion”. The industry bluntly said: “The spot price of DDR4 is actually higher than that of the new generation DDR5. This phenomenon has not been seen for ten years.”

Since June alone, the spot price of DDR4 8Gb (1G×8) 3200 has risen from an average price of $2.73 on May 30 to the current $3.775, a surge of 38.27%. Since the beginning of this quarter, it has risen from $1.63 on March 31, and has soared nearly 1.32 times.

Samsung and Micron will stop supplying DDR4 one after another. The market is worried about the subsequent supply shortage, which has triggered a desire to buy at a high price. Mysterious buyers have made large-scale purchases of DDR4, which indicates that OEM/ODM manufacturers are making every effort to consolidate the supply of DDR4.

Analysts believe that as manufacturers’ technology becomes more advanced and depreciation decreases, the current spot price of DDR4 has not only crossed the break-even point of DRAM manufacturers, but the profits are also more attractive.

The post Samsung and Micron have successively stopped production of DDR4, and the price has surpassed DDR5 appeared first on INCUS IT - IT Hardware Distributor.

It is reported that DGX Spark is Nvidia’s smallest AI supercomputer to date. Currently, it is only known that the corresponding product is equipped with a GB10 Grace Blackwell chip, integrated with a Blackwell GPU, equipped with the fifth-generation Tensor Core and supports FP4, and can provide AI reasoning and fine-tuning computing power of up to 100 trillion operations per second (TOPS).

Interestingly, NVIDIA has also opened the DGX Spark platform to AIB manufacturers, who will launch their own customized versions. Currently, MSI and Gigabyte are known to be the main partners, and these two manufacturers have already demonstrated their DGX Spark products at Computex 2025.

Among them, MSI’s version is called EdgeXpert MS-C931, equipped with 128 GB LPDDR5X RAM, with a simple black and gray color scheme, and equipped with HDMI and Connect-X interfaces, supporting multiple connection methods.

Gigabyte’s version is called AI TOP ATOM, and its design style is similar to MSI’s (due to the fixed core architecture, each company has little room for adjustment). The machine uses Micron’s LPDDR5X RAM memory chip, and it is said that a single system can support large AI models with up to 70 billion parameters.

The post Nvidia’s first consumer AI supercomputer, DGX Spark, is expected to be launched before next month, with a starting price of about $4,000 appeared first on INCUS IT - IT Hardware Distributor.

A new report claims that Nvidia has recently raised the official prices of nearly all of its products to combat the impact of tariffs and surging manufacturing costs on its business, with gaming graphics cards receiving a 5 to 10% hike while AI GPUs see up to a 15% increase.

As reported by Digitimes Taiwan (translated), Nvidia is facing “multiple crises,” including a $5.5 billion hit to its quarterly earnings over export restrictions on AI chips, including a ban on sales of its H20 chips to China.

Digitimes reports that CEO Jensen Huang has been “shuttling back and forth” between the US and China to minimize the impact of tariffs, and that “in order to maintain stable profitability,” Nvidia has reportedly recently raised official prices for almost all its products, allowing its partners to increase prices accordingly.

Despite the hikes, Digitimes claims Nvidia’s financial report at the end of the month “should be within financial forecasts and deliver excellent profit results,” driven by strong demand for AI chips outside of China and the expanding spending from cloud service providers.

The report states that Nvidia has applied official price hikes to numerous products to keep its earnings stable, with partners following suit. As an example, Digitimes cites the RTX 5090, bought at premium prices upon release without hesitation, such that channel pricing “quickly doubled.”

The report notes that following the AI chip ban, RTX 5090 prices climbed further still, surging overnight from around NT$90,000 to NT$100,000, with other RTX 50 series cards also increasing by 5-10%. Digitimes notes Nvidia has also raised the price of its H200 and B200 chips, with server vendors increasing prices by up to 15% accordingly.

According to the publication’s supply chain sources, price hikes have been exacerbated by the shift of Blackwell chip production to TSMC’s US plant, which has driven a significant rise in the price of production, materials, and logistics.

There is some hope that the measures could be temporary; however, following the news that the US and China have agreed on a trade deal that should cut tariffs by 115%, thanks to a 90-day pause on reciprocal tariffs. As with latent price increases being passed on to consumers, however, it could be some time before prices start to fall.

The post Nvidia reportedly raises GPU prices by 10-15% as manufacturing costs surge — tariffs and TSMC price hikes filter down to retailers appeared first on INCUS IT - IT Hardware Distributor.