Samsung Electronics' next-generation 12-layer HBM4E high-bandwidth memory (HBM) product has reportedly cleared customer quality validation from Nvidia and major hyperscale data center operators. The source frames this as a report and states that, if confirmed, the result would strengthen Samsung's bid to supply memory for next-generation AI accelerators; however, order timing and mass production plans remain unclear. The source does not state the cause.
According to a TrendForce post dated Oct 9, 2026, NVIDIA's NVHBM could rewrite the HBM playbook, accelerating the custom HBM era and giving base dies a much bigger role. Under this trend, memory makers may increasingly turn to external foundries for base die production. Micron and Samsung are already rethinking their strategies, putting foundry giant TSMC increasingly in the spotlight. The source uses hedged wording ('may', 'could'), so this is an analytical outlook rather than a confirmed capacity change; the source does not state the cause beyond the NVHBM-driven custom HBM shift.
HBM base diecustom HBMMicron, Samsung and 3 totalTrendForce Official (Market Data)
According to preliminary IDC data, worldwide PC shipments fell 20.1% year over year in Q3 2026 to 62.7 million units, down from 78.5 million in the same quarter last year. It is the second straight quarterly decline and much deeper than Q2, which IDC now lists as a 3.8% drop; shipments also slipped 9.1% versus Q2, even though Q3 is normally the bigger quarter. The source attributes this to the first half of the year, when vendors and channels stocked up ahead of memory-driven price hikes, pulling volume out of the second half, and with supply still constrained and prices high, Q3 apparently had little demand left to work with. The three largest vendors were hit hardest: HP shipments dropped 30.9% to 10.3 million units, Dell fell 25% to 7.6 million, and Lenovo, still first with a 23.8% share, was down 22.6% to 14.9 million; together they gave up 4.2 points of market share. Apple and ASUS also shipped fewer PCs, down 11.3% and 8.6% respectively, but both fell slower than the market and ended up with a larger share, at 9.5% and 8.7%.
AMD has detailed the IBS Memory Profiler, a little-discussed hardware feature in its upcoming Zen 6 processors. AMD engineer Bharata Bhasker Rao presented it at the Linux Plumbers Conference (LPC) 2026 in Prague on October 5. Instruction-Based Sampling (IBS) is AMD's existing profiling hardware, and Zen 6 adds a separate, lighter instance that only samples data memory accesses; each sample hands the kernel the physical address of the access and shows whether the data came from DRAM or external memory such as CXL. AMD's EPYC 9006 "Venice" chips support memory expansion via CXL 3.1. AMD wants to feed the profiler's samples into pghot, a proposed kernel subsystem for hot page tracking; the pghot patches are still in development. The results are a mixed bag: on a test system with 256 logical CPUs and 256 GB of CXL memory, XSBench ran at 4.57x over a no-tiering baseline against 1.18x for the current method, and Graph500 hit 3.17x against 2.34x; Redis saw no gain and NAS BT landed slightly behind. Per-folio promotion cost fell from 26.1 to 15.8 microseconds in an overhead test, though that benchmark finished about 5% slower. These are AMD's own numbers, and the slides list no virtualization support for the profiler. EPYC Venice is in production, with cloud deployments expected later this year.
Zen 6 processorsIBS Memory ProfilerEPYC 9006 "Venice"AMDTechPowerUp
According to VideoCardz, citing a source on Board Channels, AMD has increased GPU vendor pricing by as much as $104 in China. The Radeon RX 9070 receives a price hike equivalent to $104, the steepest increase on the list, while the Radeon RX 9060 XT 8 GB gets the smallest increase at $45; the Radeon RX 9070 XT and RX6500 XT pricing remain unchanged. The price hikes are reportedly motivated by higher memory prices and manufacturing costs. The increases are seemingly local to China for the time being, but international board partners have also been informed of the upcoming changes.
GlobalFoundries and TSMC have signed an effective five-year, $2 billion manufacturing agreement under which GlobalFoundries will expand capacity at its Malta, NY fab to produce silicon interposers for TSMC's CoWoS advanced packaging technology, with volume production slated to ramp up in the first half of 2028.
On October 8, 2026, TSMC announced a $2 billion partnership with GlobalFoundries to have GlobalFoundries produce silicon interposers for TSMC's CoWoS advanced packaging specification at its Malta, New York fab, aiming to bolster US silicon interposer production and close a domestic supply chain gap. The partnership is expected to last five years with an option to extend, but volume production is not slated to begin until the first half of 2028.
Intel Foundry published a two-article series on chip power delivery. The first article cites Gartner forecasts that worldwide data center power demand will rise from 104 gigawatts in 2025 to 132 gigawatts in 2026 and reach 290 gigawatts by 2030; these figures are forecasts. It also states that AI-optimized servers are expected to account for 31% of consumption in 2026 and surpass conventional servers by 2027. The article explains that conventional frontside power delivery routes electricity and signals through shared metal layers, creating resistance and inductance issues as designs grow more complex; Intel says its RibbonFET Gate-all-around (GAA) transistor architecture with Intel 18A separates these functions by moving power routing to the chip's backside, reducing resistance, inductance, and IR drop. The article also describes local energy storage options such as Omni MIM on-die capacitors, substrate-embedded capacitors, and embedded deep trench capacitors, as well as implementations including fully integrated voltage regulators with CoaxMIL magnetic inductors, C2VR capacitive voltage regulators, and dedicated regulator chiplets. The second article extends the problem to chiplets and stacked dies, noting that advanced packaging can create electrical bottlenecks, and advocates moving voltage conversion closer to the load.
power deliveryRibbonFET Gate-all-around (GAA) transistorOmni MIM on-die capacitorsIntel FoundrySemiWiki
AMD CEO Lisa Su visited Samsung on Oct 7 to discuss HBM supply and foundry orders. Samsung seeks foundry business in exchange, but AMD officials cited high risk in outsourcing to Samsung. Neither party has confirmed the deal.
Samsung is reportedly planning to exit the automotive application processor (AP) business it has pursued since 2018. According to the report citing industry sources, once Samsung completes production under existing supply commitments to Hyundai Motor (agreement secured in 2023) and Germany's BMW (secured last year), it plans to discontinue the development and mass production of successor Exynos Auto products; this is a reported plan, not a confirmed halt. Samsung will focus its System LSI operations on image sensors and next-generation mobile APs as it seeks to return the division to full-year profit next year. The stated causes are high development costs from stringent automotive safety and durability requirements and Qualcomm's effective dominance of the automotive AP market, which makes it difficult for Samsung to gain share. The System LSI and foundry businesses have together posted annual losses totaling trillions of won. In its latest detailed update released with second-quarter results in July, Samsung said expanded sales of mobile SoCs and image sensors helped System LSI post record first-half revenue in 2026. Samsung is developing the Exynos 2700 mobile AP, scheduled to enter mass production in the fourth quarter on Samsung Foundry's second-generation 2nm process, and seeks to expand its adoption in the Galaxy S27 series due in 2027; Samsung spent KRW 7.44 trillion on mobile APs from external suppliers in the first half of 2026. Samsung also plans to expand businesses including display driver ICs (DDIs), power management ICs (PMICs), and chip design for data centers.
automotive application processors (APs)image sensorsnext-generation mobile APsSamsung, Samsung Electronics System LSI division and 6 totalTrendForceEOL/Discontinuation
Related manufacturers
Samsung, Samsung Electronics System LSI division, Qualcomm, Hyundai Motor, BMW, Apple
Micron and Samsung are shifting HBM base die production to TSMC, with Micron projecting higher ASP and margins for NVHBM. This follows NVIDIA's NVHBM architecture which increases base die value. Bullish: Outsourcing to TSMC and higher ASP expectations for custom HBM.
HBMHBM4HBM4EMicron, Samsung and 6 totalTrendForcePartnership/M&A
Kioxia has launched its first E1.L SSD, the LD4 Series, utilizing BiCS FLASH Gen 8 QLC flash. Targeted at hyperscale data centers and read-intensive applications, this product aims to increase storage density in 1U servers.
According to a report by Chosun.biz (published 8 October 2026), AMD CEO Lisa Su is visiting South Korea to secure additional supply of HBM memory for AI accelerators, saying 'We are going to talk about our broader partnership' with Samsung executives; Samsung, meanwhile, reportedly 'hopes to turn its memory supply relationship with AMD into foundry' orders for logic chips. The visit comes a little less than seven months after the two companies signed a non-binding memorandum of understanding (MOU) this March, effective for three years and covering four areas: supply of next-generation memory products including HBM4 and advanced DDR5; co-development and optimization of memory solutions for AI and high-performance computing systems; exploration of foundry manufacturing opportunities for future AMD platforms; and co-optimization of advanced packaging and related semiconductor technologies. Samsung has become a principal supplier of HBM4 for AMD's Instinct MI400-series accelerators; expanding collaboration beyond memory supply is considerably more complicated, as it involves considerable efforts to design chips for Samsung Foundry's process technologies and risks associated with a product ramp at a new contract chipmaker — AMD's most recent Annual Report states that transitioning production of some products to new manufacturers may cause delayed product introductions, lower yields or poorer performance. AMD has already started shipments of its Instinct MI455 accelerators with 432GB of HBM4 onboard as well as rack-scale Helios AI systems to interested parties; Su said 'Samsung Electronics and SK hynix are key supply-chain partners for AMD,' that 'AI demand is extremely high' so AMD is planning its long-term supply chain with partners, and that 'It is important to plan three to five years ahead with our partners.' The article concludes that AMD is negotiating for more memory supply and whether it will take the risk of adopting Samsung Foundry beyond niche programmable chips 'only time will tell.'
HBM4DDR5HBM memoryAMD, Samsung and 5 totalTom's HardwarePartnership/M&A
In an analysis published October 08, 2026, TrendForce said the biggest challenge for the notebook market next year is no longer getting hold of components but whether consumers are willing to pay for them. TrendForce currently expects global notebook shipments to fall by a low-single-digit percentage in 2027; if DRAM and CPU prices remain elevated and brands pass more of the cost on to consumers, higher retail prices could lengthen replacement cycles further and push the decline into the high-single-digit range. 2026 shipments were supported by improved CPU availability, early procurement by brands and replacement purchases brought forward, which TrendForce does not see as a real recovery in end demand, as some purchases originally expected in the second half of 2026 and beyond have already been made. In a mainstream notebook with a suggested retail price of USD 900, the CPU, DRAM and SSD accounted for around 68% of BOM cost in the third quarter of 2026. On the component side, DRAM is likely to remain tight as AI servers continue to absorb advanced process capacity while PC DRAM supply growth risks remaining limited; NAND Flash constraints could ease in the second half of 2027 as new capacity comes online; and PCBs, MLCCs and PMICs are under pressure from capacity allocation, raw material costs and geopolitical factors, which could mean higher prices or longer lead times. TrendForce also estimates the share of global notebook production outside China will fall from around 24% in 2025 to 21% in 2026 and could drop below 20% in 2027.
Microsoft opened pre-orders for the Surface Laptop Ultra on October 7, a model featuring up to 128GB of unified memory, highlighting a potential new source of dynamic random-access memory (DRAM) demand as PC makers bring larger AI models and autonomous agents onto local devices. But the opportunity comes as PC shipments decline and memory manufacturers prioritize server customers, raising questions about whether high-capacity PCs can secure enough supply to become a meaningful growth market.
Samsung's preliminary Q3 operating profit is KRW 107.4 trillion, up 782.5% YoY. If confirmed, it would be the first Korean company to top KRW 100 trillion in a quarter. Q3 revenue was KRW 195 trillion, with full results due this month. The source does not state the cause of the profit increase.
Team Group, a Taiwanese company, reported a sharp rise in September 2026 revenue, up 16.17% from August and 109.69% year-on-year, with revenue of NT$4.081 billion (US$128.1 million), marking its second-highest monthly total on record. The increase occurred amid a global memory upswing, with strong demand across segments and capacity shifts toward AI-related infrastructure continuing to reshape the market. The memory upswing may affect pricing, supply, and procurement decisions for server operators, device makers, and distributors worldwide. The source does not state the cause of the revenue increase.
Microsoft released its first Surface notebook powered by Nvidia's RTX Spark SoC (the N1X chip), the Surface Laptop Ultra, during an event in San Francisco on Wednesday. It is one of several new Copilot+ PCs aimed at 'builders,' available for preorder starting today and expected to begin shipping next week. It starts at $2,599 for the gray variant with the lower-end 18-core N1X, 24 GB of unified memory, and a 512 GB user-serviceable SSD; stepping up to the 20-core chip, 128 GB of memory, and 1 TB of storage costs $5,899. Microsoft also announced a mini PC style development kit based on the N1X priced at $5,999. The machine ships with either an 18- or 20-core CPU designed by MediaTek and Arm and an Nvidia Blackwell-based RTX GPU with between 5,120 and 6,144 cores, with up to 128 GB of memory. The source does not state any impact on supply or pricing.
Samsung Electronics reported a record 783% YoY operating profit surge to 107.4 trillion KRW in Q3 2026, driven by HBM4 demand and memory price gains. HBM bit shipments jumped nearly 50% QoQ, while DRAM prices rose amid tight supply. Bullish: Supply constraints and price uptrends are confirmed.
Rambus Inc. issued a press announcement ahead of the recent AI Infra Summit and was present at the summit to spotlight its new production-ready CryptoManager Root of Trust supporting the Caliptra specification. The solution introduces hardware-level security orchestration built directly around the unmodified open-source Caliptra core; by integrating a secure RISC-V processor, protected memory, and dedicated cryptographic accelerators alongside specialized drivers, it expands security visibility across the entire System-on-Chip (SoC). The stated cause is the gap between an open-source specification and production-ready, commercially deployable silicon: an open core typically lacks platform-wide security orchestration, and chip vendors otherwise shoulder all integration and schedule risks alone. The offering covers physical attack defense (side-channel analysis and fault injection countermeasures), cryptographic agility blending classical algorithms like AES, SHA, and RSA with emerging Post-Quantum Cryptography standards such as ML-KEM and ML-DSA, and readiness for certifications including FIPS 140-3, PSA, or SESIP.