New sales data shared by The Game Business suggests physical PS5 games still have plenty of buyers. According to Ampere Analysis, 35% of worldwide sales of Marvel's Wolverine in September came from boxed copies. Ampere estimates the game sold around 2.16 million copies in its launch month, which would put physical sales at roughly 756,000. Interestingly, Europe leaned even harder towards discs, with GSD data showing 51% of sales there coming from the boxed version. The split varies a lot by country, though. Two thirds of Wolverine's UK sales were digital, while in France more than two thirds were physical. Wolverine wasn't alone either, as over 45% of Silent Hill: Townfall sales were boxed, while The Blood of Dawnwalker on PS5 and Onimusha came in at 36% and 30%, respectively.

These numbers are interesting because Sony plans to stop producing discs for new PlayStation games from January 2028, and earlier data from Alinea Analytics pointed in a similar direction for Sony's biggest exclusives. Of course, these are estimates, and actual data available to Sony may be different. Moreover, these patterns can be affected heavily by factors such as pricing and availability. There have also been rumors that Sony could push its digital-only plans back, while Microsoft is asking Xbox Insiders whether its next console should support discs. For now, though, Sony's own messaging still points to a January 2028 cutoff.

SemiAnalysis has published a teardown of Intel's Panther Lake (Core Ultra 7 365), offering a rare look at 18A inside a shipping chip. Intel's 18A node brings advanced manufacturing technologies such as rear-side power delivery and a new transistor design, so its success is vital for the Foundry. The firm found the 18A compute logic to be similar in density to the TSMC N3E logic used in the 12-core Xe3 GPU tile. It also puts 18A about 18.6% denser than the Intel 3 GPU logic, using a representative-cell model. However, SemiAnalysis says 18A does not lead TSMC N3P, N2, or Samsung SF2 in peak density. It further cautions that whole-die density depends on cell mix and placement.

Panther Lake is Intel's first chip with RibbonFET gate-all-around transistors, and the first commercial implementation of PowerVia backside power delivery. SemiAnalysis counted four stacked ribbons per transistor, compared to three sheets on Samsung's SF2. Moreover, the Cougar Cove P-core takes roughly the same area as Lunar Lake's Lion Cove, even though L2 grew from 2.5 MiB to 3 MiB. Panther Lake's NPU 5 reportedly requires 36.9% less chip area than Lunar Lake's NPU 4, though the number of INT8 MAC units remains the same. SemiAnalysis describes the CPU cores as incremental updates.

Intel's latest "Panther Lake" SoC, featuring a compute tile built on the company's 18A node, has reportedly achieved an 80% yield. This is great news for Intel Foundry, as the 18A yields have been steadily improving. The "Panther Lake" compute die measures approximately 8.004 x 14.288 mm, resulting in a silicon area of 114.304 mm². With this design, yields are around 80%, indicating not only low defect rates for the node but also a high parametric yield, meaning a significant number of dies have all functioning elements. This suggests that Intel is overcoming its historic parametric yield problems alongside manufacturing challenges.

A testament to this progress might be the recently rumored shift in "Nova Lake" production. Intel plans to produce the majority of "Nova Lake" compute tiles in-house, with about 80-90% being manufactured by Intel Foundry and the remainder by TSMC, as part of a dual-sourcing strategy. For a mature node like 18A, defect rates are now low, around D0=0.1 or D0=0.2. It is likely that Intel has achieved the lower end of this range due to advancements in the node and months of high-volume production with "Panther Lake." Combined with the relatively small compute tile used in this SoC, parametric yields appear to be strong.

Data shared by analyst Rhys Elliott of Alinea Analytics suggests physical discs still account for a meaningful share of revenue for PlayStation's biggest exclusives, even as Sony winds down disc production starting in 2028. Marvel's Spider-Man 2 pulled in $1.2 billion overall, with 35% of that coming from discs. Ghost of Yotei sits close behind at nearly 40% of its $400 million in revenue coming from discs. Astro Bot, Stellar Blade, and Death Stranding 2 all land in a similar 32% to 47% range. Helldivers 2 is the clear outlier, with physical copies covering just 8.3% of its $235 million in revenue, but that's to be expected from a multiplayer game.

As the numbers suggest, Sony's big-name exclusives still have a sizeable audience that leans towards physical discs. Meanwhile, Sony seems to be getting more careful about its bets. Concord's cancellation reportedly cost over $300 million, and titles like Marathon, Saros, Marvel Tokon, and Lego Horizon Adventures all landed softer than hoped. Marvel's Wolverine, one of the studio's biggest remaining tentpoles, just settled at 78 on Metacritic after its review embargo lifted, matching last year's Days Gone Remaster and falling short of prestige expectations.

Today marks the day we see AMD's next-generation RDNA 5 GPU IP inside working silicon, but not in the form you might expect. Thanks to SemiAnalysis's die annotation of the Exynos 2700 SoC from Samsung, we see AMD's new GPU IP shine in a mobile form with the Xclipse 970 GPU. As earlier leaks reported, AMD and Samsung co-developed the Xclipse 970 GPU using AMD RDNA technology and adapted it for mobile. However, the GPU IP has now been upgraded to the latest RDNA 5, making this SoC technically the first RDNA 5-equipped chip that will be shipping. Built on Samsung Foundry's SF2 node, Samsung has designed this SoC with a unique 10-CPU core configuration and a strong GPU to compete with Qualcomm Snapdragon Elite SoCs.

On the CPU front, Samsung uses Arm's off-the-shelf CPU core IP, which includes C2-Ultra and C2-Pro. This setup features a single C2-Ultra big core running at 4.24 GHz, one C2-Ultra running at 3.36 GHz, four C2-Pro cores running at 3.74 GHz, and four C2-Pro cores running at 2.88 GHz, making a total of 10 CPU cores. On the GPU side, the Xclipse 970 packs eight GPU WGPs, each containing two Compute Units. This results in 16 CUs on the GPU front if everything remains the same with RDNA 5 and no changes to the WGP/CU ratio have been made by AMD since the last-generation RDNA 4.

Chinese memory maker CXMT has debuted on the Shanghai A-Class STAR stock market with one of the biggest public offerings in China. Today, CXMT's Initial Public Offering (IPO) saw the price of CXM jump from CNY 8.66 ($1.28). At the time of writing, the stock, under the ticker "688825.SS," is trading at CNY 49.00 ($7.24), indicating a 465.82% gain on its first trading day. Initial projections estimated CXMT's market capitalization at CNY 579 billion, or $78.11 billion, but the valuation has increased significantly, now reaching $513.75 billion. This makes it the world's 24th most valuable company by market capitalization, surpassing Tencent, Mastercard, and even the technology giant Intel.

This isn't a surprise, as CXMT will quickly generate revenue that will far exceed its IPO price, hence the valuation just got stronger. Some analysts, including those from SemiAnalysis, estimate that CXMT will generate about $50 billion for the entirety of 2026. In the first quarter of 2026 alone, CXMT reported $7.3 billion, marking approximately 700% growth compared to the same period a year ago, with operating margins around 70%. Massive quarterly growth, new capacity coming online, and an unprecedented demand for DRAM have pushed CXMT's valuation nearly five times higher than originally planned.

Marathon launched almost exactly three weeks ago, with a mix of both fanfare and criticism, but given its reasonably strong player counts—with a stable 28,000-40,000 daily concurrent players—it's not surprising to learn that the game has sold an estimated 1.2 million units across the three platforms where it is available. According to Alinea Analytics, Marathon has earned $55 million in revenue across PC, PlayStation, and Xbox platforms. The analytics firm estimates that Steam holds around 800,000 of those players, while Marathon sold 217,000 units on PlayStation 5 and 133,000 on Xbox Series platforms.

The researchers also looked into the correlation between Bungie's previous games and Marathon, and it seems as though the developer's reputation may have been a substantial driving force for Marathon players. As many as 78.2% of Marathon's Steam players had previously played Destiny 2, while 62.7% had played Helldivers 2, and 55.9% played Arc Raiders—although it should be noted that all three titles had colossal player counts, so some of this may be pure coincidence, and some of it could also be attributed to Steam's recommendation algorithms, which would be more likely to show Marathon to fans of similar Bungie games. There have also been rumors of Marathon potentially launching on last-gen Xbox and PlayStation consoles, but nothing has been confirmed regarding that release just yet.

According to SemiAnalysis, which recently released its InferenceX benchmark comparing the inference performance of the latest AI accelerators, AMD's next-generation Instinct MI455X accelerators, designed to compete with NVIDIA's "Vera Rubin," are facing serious manufacturing difficulties, causing delays in AMD's roadmap execution. Only low-volume production will occur this year, while customers will be generating AI tokens in Q2 2027. This means that NVIDIA's "Vera Rubin" VR200 rack-scale system will be the most powerful system shipping this summer, without competition. As AMD's first rack-scale system, the "Helios" AI rack with the MI455X UALoE72 configuration packs 72 GPUs and 6th Gen AMD EPYC "Venice" CPUs in a heterogeneous system designed to run as a single homogeneous GPU, delivering massive AI performance with multi-ExaFLOP compute output.

SemiAnalysisEngineering samples and low volume production of AMD's first rack scale MI455X UALoE72 system will be in H2 2026 while due to manufacturing delays, the mass production ramp and first production tokens will only be generated on an MI455X UALoE72 by Q2 2027.

A report put together by Deutsche Bank analysts has mentioned an ongoing supply-demand imbalance within the semiconductor market. According to Jukan's summary of the financial institute's findings, TSMC's "share of the advanced-node foundry market" could fall from an impressive 95% level, down to a mere 90% in the coming years. The industry-leading chipmaker continues to enjoy a prime position, but its international operation footprint seemingly cannot keep up with requests—mainly from fabless customers—during the ongoing AI hardware boom. Deutsche Bank believes that the Taiwanese foundry business has an order backlog extending way into 2027, with many clients requiring 3 nm wafer products. TSMC reportedly aims to get 3 nm node production lines pumping out around 190,000 wafers per month, by the end of this year. Even if this major target is met, demand is expected to outweigh supply. AI chip production-centric bottlenecks—mostly involving CoWoS packaging—have been dealt with in the recent past, but industry observers posit that VIP customers are considering alternate providers.

Allegedly, the next best options are the Samsung and Intel Foundries, despite rumors of teething problems—both companies will very likely welcome new requests. In modern times, many of Samsung's big clients have switched over to TSMC, due to (the former's) supposedly troubled development of more advanced nodes. Deutsche Bank analysts have outlined a "big six" group of key customers: AMD, Apple, Broadcom, MediaTek, NVIDIA, and Qualcomm. The fabless firms could be pursuing alternate "production capacity" avenues, with Samsung's 2 nm (SF2) and 4 nm node process lines being of particular interest. Like TSMC, the South Korea megacorp's state-of-the-art fabrication sites and staff expertise are concentrated in home territories. A surprise prediction has highlighted Samsung's main North American foundry—the Deutsche Bank report pinpointed a very specific location: "for customers seeking alternative supply, Samsung's Taylor fab is more likely to become the first destination." In addition, analysts think that the big six will select Samsung's Texas-based foundry over Intel's best native site (perhaps dealing in 18A).

Apple launched its A19 Pro and A19 mobile processors about three months ago. The 3 nm 64-bit Arm-based SoCs drive a new wave of iPhone 17 series smartphones, as well as the ultraslim + lightweight iPhone Air. It took a while for close-up die shots to emerge from independent groups. Eventually, close-ups of the A19 Pro's layout turned up online, mid-way through November. Investigations have focused on Apple's selection of TSMC's cutting-edge N3P node process; granting a noticeable reduction in die areas, when compared to previous-gen equivalents. 2024's A18 Pro and A18 designs are 3 nm chips, but manufactured via N3E (FinFET)—an older/less advanced node. The leading foundry's N3P product is referred to as a "high-performance variant."

When comparing the current flagship chipset (A19 Pro) to its predecessor (A18 Pro), keen observers noted a 10% reduction in footprint—going from 105 mm² down to 98.6 mm². Additionally, SemiAnalysis reckons that the A19 is 9% smaller than its forebear (A18). According to their calculations, the upgrade to N3P would only grant about a 4% difference in area. Logically, Apple's engineering team must have played around with other aspects in order to reach nine to ten percent shrinkage. SemiAnalysis outlined important improvements across the latest chip designs: "looking at core areas, The P-Core shrank by 4%, while the E-Core and GPU are 10% larger, reflecting the increased transistor budget for the large efficiency gains seen in benchmarks...The cache macro doubled in size to 32 KB for a 10% higher density cache: A18 4 MB of SLC: 1.08 mm², and A19 4 MB of SLC: 0.98 mm². There's also a more area efficient layout of all the minor 'uncore' SOC area (e.g. display/media engine, ISP, security etc)." In conclusion, they propose that several smart design innovations have lead to "an area reduction comparable to a major node shrink," overall a big achievement, albeit in a smaller package.

Unveiling what it describes as the most capable model series yet for professional knowledge work, OpenAI launched GPT-5.2 today. The model was trained and deployed on NVIDIA infrastructure, including NVIDIA Hopper and GB200 NVL72 systems. It's the latest example of how leading AI builders train and deploy at scale on NVIDIA's full-stack AI infrastructure.

Pretraining: The Bedrock of Intelligence
AI models are getting more capable thanks to three scaling laws: pretraining, post-training and test-time scaling. Reasoning models, which apply compute during inference to tackle complex queries, using multiple networks working together, are now everywhere. But pre-training and post-training remain the bedrock of intelligence. They're core to making reasoning models smarter and more useful. And getting there takes scale. Training frontier models from scratch isn't a small job. It takes tens of thousands, even hundreds of thousands, of GPUs working together effectively. That level of scale demands excellence across many dimensions. It requires world-class accelerators, advanced networking across scale-up, scale-out and increasingly scale-across architectures, plus a fully optimized software stack. In short, a purpose-built infrastructure platform built to deliver performance at scale.

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.

While AMD hasn't launched its upcoming Instinct MI400 series of accelerators yet, the company has started preparing its supply chain for what comes after. Thanks to SemiAnalysis, we have a clearer picture of what AMD's Instinct MI500 UAL256 rack will look like. At the base of each compute node, there will be a single "Verano" CPU paired with four Instinct MI500 accelerators. With up to 64 nodes, split across two 32-node racks, AMD plans to enable 256 logical GPUs and 64 CPUs per mega pod, working in tandem for advanced AI and HPC workloads. Instinct MI500s UAL256 will be AMD's second-generation rack-scale system, following the upcoming Instinct MI450X IF128 aka "Helios", scheduled for the second half of 2026. The Instinct MI500 UAL256 scale-up mega pod is scheduled to arrive in 2027, just in time for Verano and Instinct MI500 accelerator.

The Instinct MI500 UAL256 scale-up mega pod splits the compute into two 32-rack cabinets, with a single central rack connecting them and hosting 18 switching trays for networking. Each of these networking trays contains four 102.4T "Vulcano" switch ASICs, designed for 800G external throughput, and manufactured on TSMC's 3 nm node. For the central processor, AMD showed few details at its Advancing AI event, which claimed that Verano introduces the "Zen 7" microarchitecture for even higher IPC and support for even newer instruction sets. For now, it's not clear if AMD will increase CPU core counts beyond the up to 256 cores per package of "Venice." Still, we're hearing that Verano will retain the Socket SP7 infrastructure, which means it will likely retain the memory and PCIe interfaces introduced in 2026, such as PCIe Gen 6, UALink, and Ultra Ethernet.

QNAP Systems, Inc., a leading computing, networking, and storage solutions innovator, announced the rollout of a major AI algorithm upgrade for DA Drive Analyzer, the AI-powered drive health prediction service jointly developed with storage testing leader ULINK Technology. The upgraded engine significantly improves the accuracy of hard drive failure predictions compared to traditional S.M.A.R.T. (Self-Monitoring, Analysis and Reporting Technology) monitoring. QNAP NAS users can now benefit from more proactive and reliable data protection.

While most users still rely on built-in S.M.A.R.T. attributes to assess drive health, S.M.A.R.T. is fundamentally a reactive mechanism and often fails to provide early warnings before a drive fails. "S.M.A.R.T. alone isn't enough—many users are caught off guard by silent drive failures," said Tim Lin, Product Manager at QNAP. "DA Drive Analyzer uses an AI model trained on behavioral data from millions of drives. It can detect potential anomalies even before any S.M.A.R.T. warning is triggered, enabling users to take preventive action and avoid data loss."

Late last month, Kurnal (@Kurnalsalts) shared a partial die shot of a supposed Nintendo Switch 2 chipset—this teaser image seemed to verify previous leaked claims about the forthcoming next-gen hybrid gaming console being powered by a custom NVIDIA "T239" SoC design. Two weeks after the fact, Kurnal has boasted about delivering an alleged "world's first Nintendo Switch 2 die shot." Their social media post included a couple of key specification data points: "Samsung 8N (8 nm), eight Cortex-A78C cores, (shared) 4 MB L2 cache, and 1536 CUDA/6TPC 'Ampere' GPU." Another leaker—Geekerwan—said that they acquired a "Switch 2 motherboard" via Xianyu. This Chinese equivalent to eBay seems to be a veritable treasure trove of tech curiosities.

Earlier on in 2025, black market sellers were attempting to offload complete pre-launch Switch 2 packages for big money. As reported by VideoCardz, recent acquisitions only involved the securing of non-functional motherboard + SoC units—Kurnal disclosed a 1000 RMB (~$138 USD) price point. Digital Foundry's Richard Leadbetter is a very visible advocate of the Switch 2 chipset being based on a mature 8 nm Samsung node process. His personal belief was aimed at certain critics; these opposers predicted 5 nm manufacturing origins. Older leaks suggested a larger than expected die footprint—relative to Switch 1's internal setup; almost twice the size—leading to Leadbetter's conclusion. Comparison charts—produced by Kurnal and Geekerwan—propose an occupied area of 207 mm².

Earlier today, TechPowerUp covered the alleged performance prowess of Huawei's CloudMatrix 384 system super node. According to SemiAnalysis opinion, the system's Ascend 910C AI accelerators are a generation behind—in terms of chip performance—when compared to NVIDIA's GB200 "Blackwell" AI GPU design. SMIC seemed to be in the picture, as Huawei's main fabrication partner—possibly with an in-progress 5 nm node process. Instead, SemiAnalysis has surmised that the Ascend 910C is based on plenty of non-native technologies. Huawei's (current and prior) "aggressive skirting of export controls" has likely enabled the new-gen AI chip's better than expected performance stats. SemiAnalysis documented the early sample's origins: "while the Ascend chip can be fabricated at SMIC, we note that this is a global chip that has HBM from Korea (Samsung), primary wafer production from TSMC (Taiwan), and is fabricated by 10s of billions of wafer fabrication equipment from the US, Netherlands, and Japan...One common misconception is that Huawei's 910C is made in China. It is entirely designed there, but China still relies heavily on foreign production."

Despite China's premiere foundry business making pleasing in-roads with a theorized "7 nm N+2" manufacturing test line, Huawei has seemingly grown impatient with native immature production options. Today's SemiAnalysis article presents a decent dose of inside knowledge: "while SMIC does have 7 nm, the vast majority of Ascend 910B and 910C are made with TSMC's 7 nm. In fact, the US Government, TechInsights, and others have acquired Ascend 910B and 910C and every single one used TSMC dies. Huawei was able to circumvent the sanctions on them against TSMC by purchasing ~$500 million of 7 nm wafers through another company, Sophgo...It is rumored Huawei continues to receive wafers from TSMC via another 3rd party firm, but we cannot verify this rumor." Another (fabless) Chinese chip design firm—Xiaomi—appears to still have direct/unrestricted access to TSMC manufacturing lines, albeit not for enterprise-grade AI products.

Huawei announced its CloudMatrix 384 system super node, which the company touts as its own domestic alternative to NVIDIA's GB200 NVL72 system, with more overall system performance but worse per-chip performance and higher power consumption. While NVIDIA's GB200 NVL72 uses 36 Grace CPUs paired with 72 "Blackwell" GB200 GPUs, the Huawei CloudMatrix 384 system employs 384 Huawei Ascend 910C accelerators to beat NVIDIA's GB200 NVL72 system. It takes roughly five times more Ascend 910C accelerators to deliver nearly twice the GB200 NVL system performance, which is not good on per-accelerator bias, but excellent on per-system level of deployment. SemiAnalysis argues that Huawei is a generation behind in chip performance but ahead of NVIDIA in scale-up system design and deployment.

When you look at individual chips, NVIDIA's GB200 NVL72 clearly outshines Huawei's Ascend 910C, delivering over three times the BF16 performance (2,500 TeraFLOPS vs. 780 TeraFLOPS), more on-chip memory (192 GB vs. 128 GB), and faster bandwidth (8 TB/s vs. 3.2 TB/s). In other words, NVIDIA has the raw power and efficiency advantage at the chip level. But flip the switch to the system level, and Huawei's CloudMatrix CM384 takes the lead. It cranks out 1.7× the overall PetaFLOPS, packs in 3.6× more total HBM capacity, and supports over five times the number of GPUs and the associated bandwidth of NVIDIA's NVL72 cluster. However, that scalability does come with a trade-off, as Huawei's setup draws nearly four times more total power. A single GB200 NVL72 draws 145 kW of power, while a single Huawei CloudMatrix 384 draws ~560 kW. So, NVIDIA is your go-to if you need peak efficiency in a single GPU. If you're building a massive AI supercluster where total throughput and interconnect speed matter most, Huawei's solution actually makes a lot of sense. Thanks to its all-to-all topology, Huawei has delivered an AI training and inference system worth purchasing. When SMIC, the maker of Huawei's chips, gets to a more advanced manufacturing node, the efficiency of these systems will also increase.

TrendForce's latest investigations reveal that ongoing advancements in OLED displays are propelling the growth of QD-OLED monitor shipments. QD-OLED's share of OLED monitor shipments is expected to rise from 68% in 2024 to 73% in 2025, highlighting its strong competitiveness in the high-end monitor market. Driven by growing market demand, more brands and product lines are expected to adopt QD-OLED to deliver superior image quality and refresh rates for gamers and professional users.

The introduction of new 27-inch UHD products and high-refresh-rate QHD 500 Hz monitors is drawing attention. These models offer strong appeal for gaming and professional use, becoming key drivers of QD-OLED monitor adoption. The penetration rate of QD-OLED in 27-inch models is projected to grow from 32% in 2024 to 47% in 2025, driven by an expanding product lineup and ongoing technical improvements.

High-performance GPUs manufactured in Taiwan could now enter the US market tariff-free through a technical loophole in the United States-Mexico-Canada Agreement (USMCA), found by a research firm SemiAnalysis. Companies can route Taiwan-made GPUs through assembly facilities in Mexico and Canada, effectively circumventing the 32% import duty that would otherwise apply to direct shipments from Taiwan. The exemption hinges on a Most-Favored-Nation clause within the USMCA framework that specifically classifies digital processing units (HTS 8471.50), automatic data processing machine units (HTS 8471.80), and their associated components (HTS 8473.30) as "originating goods." This classification applies regardless of manufacturing origin, creating a duty-free pathway for NVIDIA HGX boards, GB200 baseboards, and RTX GPU cards that undergo final assembly in North American facilities.

The strategy capitalizes on two complementary policy mechanisms. First, President Trump's March 7 executive orders maintained existing USMCA exemptions, preserving the duty-free status for compliant goods from Canada and Mexico. Second, the USMCA's expanded definition of originating products creates a classification framework that treats assembled servers and related components as North American products despite their core manufacturing in Taiwan. For US technology firms, the additional logistical complexity of cross-border assembly operations is offset by eliminating substantial import duties on these high-value components. This practice mirrors established protocols in agricultural imports, where products like Mexican avocados gain preferential treatment under similar origin rules. The global supply chain is adapting quickly, especially in high-margin areas like GPUs, which power AI workloads. We are yet to see how companies set up manufacturing and logistics in the new era of tariff-driven narrative.

According to a report produced by semiconductor industry analysts at Kiwoom Securities—a South Korean financial services firm—Semiconductor Manufacturing International Corporation (SMIC) is expected to complete the development of a 5 nm process at some point in 2025. Jukanlosreve summarized this projection in a recent social media post. SMIC is often considered to be China's flagship foundry business; the partially state-owned organization seems to heavily involved in the production of (rumored) next-gen Huawei Ascend 910 AI accelerators. SMIC foundry employees have reportedly struggled to break beyond a 7 nm manufacturing barrier, due to lack of readily accessible cutting-edge EUV equipment. As covered on TechPowerUp last month, leading lights within China's semiconductor industry are (allegedly) developing lithography solutions for cutting-edge 5 nm and 3 nm wafer production.

Huawei is reportedly evaluating an in-house developed laser-induced discharge plasma (LDP)-based machine, but finalized equipment will not be ready until 2026—at least for mass production purposes. Jukanlosreve's short interpretation of Kiwoom's report reads as follows: (SMIC) achieved mass production of the 7 nm (N+2) process without EUV and completed the development of the 5 nm process to support the mass production of the Huawei Ascend 910C. The cost of SMIC's 5 nm process is 40-50% higher than TSMC's, and its yield is roughly one-third." The nation's foundries are reliant on older ASML equipment, thus are unable to produce products that can compete with the advanced (volume and quality) output of "global" TSMC and Samsung chip manufacturing facilities. The fresh unveiling of SiCarrier's Color Mountain series has signalled a promising new era for China's foundry industry.

The global semiconductor manufacturing industry closed 2024 with strong fourth quarter results and solid year-on-year (YoY) growth across most of the key industry segments, SEMI announced today in its Q4 2024 publication of the Semiconductor Manufacturing Monitor (SMM) Report, prepared in partnership with TechInsights. The industry outlook is cautiously optimistic at the start of 2025 as seasonality and macroeconomic uncertainty may impede near-term growth despite momentum from strong investments related to AI applications.

After declining in the first half of 2024, electronics sales bounced back later in the year resulting in a 2% annual increase. Electronics sales grew 4% YoY in Q4 2024 and are expected to see a 1% YoY increase in Q1 2025 impacted by seasonality. Integrated circuit (IC) sales rose by 29% YoY in Q4 2024 and continued growth is expected in Q1 2025 with a 23% increase YoY as AI-fueled demand continues boosting shipments of high-performance computing (HPC) and datacenter memory chips.

Last week, Samsung introduced its brand-new Galaxy S25 smartphone series—press material focused largely on various implementations of AI features, but industry watchdogs noted the crucial selection of Qualcomm-designed processors. A "first-of-its-kind customized Snapdragon 8 Elite Mobile Platform for Galaxy chipset" is the natural choice for this generation of Samsung flagship phones, given that proprietary Exynos designs have reportedly missed the mark (yet again). Samik Chatterjee—a J.P. Morgan analyst—believes that the latest collaboration will swell Qualcomm's revenues; he predicts a gain of $2 billion (USD). Snapdragon 8 Elite processors are utilized by the entire range of globally-released Galaxy S25 models.

The previous-gen S24 series featured a mix of Qualcomm-designed chips and Samsung Exynos silicon (for different regional markets)—market analysis estimates a total of 40 million unit shipments back in 2024. Previously, Qualcomm had a 70% share of Galaxy S24 chipsets—fast-forwarding to the present day, it becomes 100% with the rollout of Galaxy S25. Late last year, press outlets posited that the South Korean company's foundry division had moved on from a "problematic" 3 nm Gate-All-Around (GAA) process. Samsung's flagship-tier Exynos 2500 SoC was linked to this node, but insiders reckon that alternative external production avenues were explored—most notably with an arch-rival: TSMC. Industry moles reckon that Samsung's leadership has slashed foundry budgets for 2025—reports from last week suggest investments being halved, as teams move onto two nanometer processes.

Hi everyone, Assassin's Creed Shadows is launching March 20, inviting you to experience the intertwined stories of Naoe, an adept shinobi Assassin, and Yasuke, a powerful African samurai. Today, you can pre-order the game on console and PC, and read up on Shadows' upcoming expansion, Claws of Awaji, which brings 10 hours of additional content free with your pre-order.

For those of you playing on PC, we've got all of Assassin's Creed Shadows' recommended PC specs listed in this article. Assassin's Creed Shadows will support raytraced global illumination and reflections, and will feature an in-game benchmark tool for performance analysis, ultra-wide resolutions, an uncapped framerate, and more. Check out all the specs chart below.

The battle of AI acceleration in the data center is, as most readers are aware, insanely competitive, with NVIDIA offering a top-tier software stack. However, AMD has tried in recent years to capture a part of the revenue that hyperscalers and OEMs are willing to spend with its Instinct MI300X accelerator lineup for AI and HPC. Despite having decent hardware, the company is not close to bridging the gap software-wise with its competitor, NVIDIA. According to the latest report from SemiAnalysis, a research and consultancy firm, they have run a five-month experiment using Instinct MI300X for training and benchmark runs. And the findings were surprising: even with better hardware, AMD's software stack, including ROCm, has massively degraded AMD's performance.

"When comparing NVIDIA's GPUs to AMD's MI300X, we found that the potential on paper advantage of the MI300X was not realized due to a lack within AMD public release software stack and the lack of testing from AMD," noted SemiAnalysis, breaking down arguments in the report further, adding that "AMD's software experience is riddled with bugs rendering out of the box training with AMD is impossible. We were hopeful that AMD could emerge as a strong competitor to NVIDIA in training workloads, but, as of today, this is unfortunately not the case. The CUDA moat has yet to be crossed by AMD due to AMD's weaker-than-expected software Quality Assurance (QA) culture and its challenging out-of-the-box experience."

Advantech, a leading provider of AIoT platforms and services, proudly unveils its latest AI acceleration modules: the EAI-1200 and EAI-3300, powered by Hailo-8 AI processors. These modules deliver AI performance of up to 52 TOPS while achieving more than 12 times the power efficiency of comparable AI modules and GPU cards. Designed in standard M.2 and PCIe form factors, the EAI-1200 and EAI-3300 can be seamlessly integrated with diverse x86 and Arm-based platforms, enabling quick upgrades of existing systems and boards to incorporate AI capabilities. With these AI acceleration modules, developers can run inference efficiently on the Hailo-8 NPU while handling application processing primarily on the CPU, optimizing resource allocation. The modules are paired with user-friendly software toolkits, including the Edge AI SDK for seamless integration with HailoRT, the Dataflow Compiler for converting existing models, and TAPPAS, which offers pre-trained application examples. These features accelerate the development of edge-based vision AI applications.

EAI-1200 M.2 AI Module: Accelerating Development for Vision AI Security
The EAI-1200 is an M.2 AI module powered by a single Hailo-8 VPU, delivering up to 26 TOPS of computing performance while consuming approximately 5 watts of power. An optional heatsink supports operation in temperatures ranging from -40 to 65°C, ensuring easy integration. This cost-effective module is especially designed to bundle with Advantech's systems and boards, such as the ARK-1221L, AIR-150, and AFE-R770, enhancing AI applications including baggage screening, workforce safety, and autonomous mobile robots (AMR).