News Posts matching #2 nm

Return to Keyword Browsing

Jensen Huang Believes that NVIDIA is TSMC's Premiere Customer - Foundry Reportedly Deprioritizes Apple

A recent Jensen Huang declaration and fresh chip production industry rumors have coalesced nigh simultaneously. During a sitdown with Jodi Shelton—on her "A Bit Personal with..." podcast—the Team Green boss made a lofty claim when asked about his confident approach when conducting business: "well, you know, it's rough to know everything—I'm just kidding...By the way, Morris (Chang) will be happy to know (that) NVIDIA is TSMC's largest customer now." Almost perfectly timed industry whispers have emerged—within the past day or two—suggesting Taiwan's leading foundry service adjusting its supply relationship with Apple. Previous to current tremors, the Cupertino-headquartered giant has reportedly enjoyed priority access to cutting-edge node processes. Yesterday morning, the "Fixed-focus Digital Cameras" Weibo account let slip about a speculated trip, stateside, made by TSMC's Che-Chia Wei.

The short blog entry alleged that the TSMC CEO visited: "Apple HQ (in order) to demand the largest price increase in recent years, and Apple may no longer enjoy priority production capacity. TSMC's largest customer is now NVIDIA, accounting for nearly 13% of total revenue." The two fabless rivals seemingly want unobstructed access to 2 nm wafer production lines, but Team Green has supposedly got the edge here—very likely driven by successes in the AI enterprise field. According to fairly fresh inside track information, Apple has secured around half of TSMC's 2 nm manufacturing capacity—linked to forthcoming A20 and A20 Pro smartphone chipsets. In terms of proprietary AI server-grade designs, the house of Mac and iPhone is reported to be working on an inference-only "Baltra" GPU project. Industry moles have mentioned the selection of a TSMC 3 nm node (N3E or N3P) for this far-off-from-launch prospect.

Samsung's Texas Fab Reportedly Attracting Major Customer Interest as AI Boom Continues

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).

Tesla Completes AI5 Chip, Revives Dojo 3 with Intel Packaging Deal

Tesla has finalized the development of its AI5 automotive processor, allowing for production at TSMC and Samsung facilities. This advancement also revives the previously halted Dojo 3 supercomputer project. CEO Elon Musk announced that the AI5 chip will offer performance on par with NVIDIA's "Hopper" architecture, with two AI5 units equaling the power of a single "Blackwell" processor. The revival of Dojo 3 project was rumored with Intel joining as a key packaging partner, marking a shift from Tesla's earlier dependence on TSMC for complete manufacturing. Intel will manage assembly and testing using its EMIB technology, which connects multiple dies via silicon bridges instead of a full-wafer interposer. This method is better suited to Tesla's large Dojo modules that combine several 654 mm² chips into a single package.

Also based on previous information, Samsung will produce the D3 training chips at its Texas facility using a 2 nm process, leaving Intel to oversee packaging operations exclusively. This division of labor addresses capacity constraints and provides Tesla with more flexibility in customizing interconnect layouts. For the AI5 automotive chips, both Samsung and TSMC will create distinct versions, but Tesla aims to ensure identical software performance across both. Initial projections suggest the AI5 could operate at just 150 W while matching the performance of NVIDIA's H100 accelerator that requires 700 W. This efficiency is achieved by removing general-purpose graphics subsystems and optimizing the architecture specifically for Tesla's neural processing.

Samsung Foundry Supposedly Achieves 50% Yields with 2 nm GAA Node Process

Late last year, conflicting insider reports provided jumbled details of Samsung's cutting-edge 2 nm GAA node process (SF2). Around that period, the South Korean megacorp's semiconductor business had initiated "mass production" of proprietary next-gen Exynos 2600 mobile processors. Earlier today, a ZDNet news article claims that yields of Samsung's first-generation 2 nm wafer product have been recorded hitting a "relatively stable" 50% milestone. This more recent achievement is puzzling, given that other inside sources disclosed a similar level being reached last September. Ideally, closer to 70% yield rates are preferred for mass manufacturing of finalized chip designs. Throughout 2025, insider reports suggested a not so smooth development of SF2—company leadership was determined to get this technology over the finishing line. Industry observers think that the alleged failure of Samsung Foundry's 3 nm line had caused potentially returning clients to stick with TSMC.

The upcoming launch of Exynos 2600 chip will be a crucial "proof of concept" moment for Samsung—with the opportunity to attract VIP customers away from a main rival's best 2 nm and 3 nm offerings. The ZDNet report mentions a second-gen 2 nm GAA option (SF2P). This refinement could be the result of "additional growth engines," and inside track info posits that guidelines have been sent to design solution partners. When compared to the first iteration, SF2P reportedly does better with: "12% improved performance, 25% improved power efficiency, and an 8% smaller footprint." One anonymous source divulged that SF2P's basic process design kit (PDK) was completed mid-way through last year. ZDNet has linked this second-gen node process to a mysterious "Exynos 2700" SoC.

TSMC Expecting to Start 3 nm Production in Arizona in 2H 2027

Almost two years after it started mass production of 2 nm chips at Fab 22 in Kaohsiung, Taiwan, TSMC is expecting to kick off 3 nm production in what will be its second fab in Arizona, USA, according to a comment by its Chairman C.C. Wei to the Taiwan media. The new fab was originally not planned to go online until 2028, but it has now been moved forward to the second half of 2027 and TSMC is also planning a fourth fab and an advanced chip assembly plant on the same plot. TSMC has pledged to invest a further US$100 billion to build a further three fabs in Arizona, alongside two IC assembly plants and an R&D centre, with 2 nm production currently slated for 2029.

The company has acquired additional land in Arizona to be able to expand its complex in Arizona into what Wei called a "megafab cluster" to be able to meet their customers' demands. What is less clear is where TSMC is going to find the factory workers, as the company has already been having issues with labour in the US. Wei also mentioned that construction of its second fab in Kumamoto, Japan has started construction and that the construction of its fab in Dresden, Germany is proceeding according to plans. At the same time, TSMC is planning additional fabs in Hsinchu and Kaohsiung that will be building chips on its 2 nm node or more advanced nodes.

MediaTek Reportedly Prioritizing Development of AI ASICs Over Mobile Chips

MediaTek's growing partnership with Google has reportedly prompted a shift in development priorities; industry insiders propose an ongoing "realignment" of internal resources. The alleged formation of a dedicated ASIC division seems to be draining staff away from the more visible Dimensity mobile chip projects—by previous announcements, a very important aspect of the Taiwanese fabless semiconductor firm's business. Going back to March 2025, MediaTek and Google partnered up on the making of a seventh-gen "Ironwood" Tensor Processing Unit (TPU) design. Later in the year, it emerged that MediaTek engineers were assisting on Google's even more futuristic "Zebrafish" TPUv8x project.

Industry analysts posit that the demands of formulating Google's complex/intricate 3 nm TPU v7 design has taken precedence within MediaTek operations; prompting the rumored formation of a full-fledged ASIC department. Another key factor seems to be a tight timeline that leads to significant volume orders of Google TPUs. According to a Commercial Times Taiwan news article, mass production is expected to start within the third quarter of 2026. Future targets include five million units pumped out by 2027, followed by an ambitious seven million unit goal in 2028. Similarly, Broadcom is "actively preparing production capacity" for its Google TPU collaboration. Local reportage points out that Rick Tsai—MediaTek's Vice Chairman and CEO—anticipates pleasing financial results related to his team's "first ASIC project." Smooth progress in 2026 could lead to an approximate $1 billion revenue contribution, with forecasts of several billion dollars in 2027. Another major earner could emerge from a speculated deal with Meta—MediaTek's second ASIC project is said to be based on an unnamed 2 nm node process.

TSMC Kicks Off 2 nm Mass Production

TSMC has announced that it has started 2 nm mass production in Fab 22 which is located in the southern city of Kaohsiung in Taiwan. This is TSMC's first fab this far south in Taiwan and it's the first phase of as many as five that will be at this location. The fab is using 300 mm wafers and TSMC's first-generation nanosheet transistor technology on the N2 node. TSMC has already stated that the N2 node is offering a full node jump in terms of performance and power benefits, although the company is comparing the N2 node to its second generation N3E node, not the original N3 node, suggesting that the N2 node is slightly better than a regular full node improvement for TSMC.

The N2 node also includes some new additions that TSMC claims will further boost performance, in the shape of a low-resistance redistribution layer (RDL) and super high-performance metal-insulator-metal (MiM) capacitors. Time will tell how these advancements play out in real world products, once we find out which customers are first out with chips built on the N2 node. The N2 node is said to offer a 10 to 15 percent performance increase at the same power consumption or a 25 to 30 percent reduction in power at the same performance, compared to the N3E node, as well as a transistor density increase of around 15 percent.

AMD "Zen 6" SKUs with 3D V-Cache Could Pack Up to 288 MB of Cache

AMD's upcoming "Zen 6" client processors are expected to continue the trend of offering optional SKUs with 3D V-Cache, similar to their predecessors. However, these new processors might feature significantly larger cache capacities compared to the current "Zen 5" SKUs in the "Granite Ridge" family. According to HXL on X, the single-CCD version will include 144 MB of 3D V-Cache, while the dual-CCD configuration will have 288 MB of cache. This suggests that AMD will maintain its familiar CPU design with a client I/O die and two separate CCDs, each benefiting from increased cache capacity.

This approach is quite similar to Intel's upcoming "Nova Lake" processors, which are also set to receive a substantial boost in cache capacity through the big Last Level Cache (bLLC). Rumors suggest that these will offer 144 MB for SKUs with a single compute die and 288 MB for those with two compute dies. Intel already employs bLLC in its "Clearwater Forest" server processors as a passive interposer, integrating local cache beneath active tiles. Including this technology in standard gaming chips could significantly enhance performance, positioning them as strong competitors against the upcoming "Zen 6" processor family.

TSMC USA's Access to Cutting-edge Nodes Could be Stymied by Taiwanese Regulations

According to recent reports, TSMC leadership is expediting a major next phase at the Arizona campus—a second fabrication facility could come online by 2027, with a move into mass production of 3 nm (N3) wafers. A fresh news article—published by CNA Taiwan—signals an uncertain technological roadmap for the leading foundry service's North American operations. Taiwanese regulators have reportedly questioned TSMC's commitment to a long-term "N-2" rule. Under strict guidelines, the nation's premier chipmaker is not allowed to narrow a two generation-wide gap between domestic and offshore foundry capabilities. Presently, the Arizona Fab 21 Phase 1 site—at best—can pump out 4 nm (N4) products. Additionally, it produces more mature 5 nm (N5) wafers.

Given the current global political climate, TSMC's executive team could have a tough time satisfying demands from both sides of the Pacific. CNA's news piece highlighted observations made by Lin Fa-cheng, the Deputy Minister of Taiwan's National Science and Technology Council (NSTC). According to that report, Fa-cheng believes that TSMC will continue to adhere to the "N-2" policy, due to the vast majority of company R&D expertise being concentrated domestically—once native development moves onto the creation of truly cutting-edge 1.2 nm and 1.4 nm technologies, their 1.6 nm-class process (A16) knowledge can be shared with overseas facilities. Almost a month ago, insiders posited that TSMC was planning another expansion of its home turf footprint—a speculated $28 billion investment could lead to the construction of three 2 nm specialist fabs.

Samsung Unveils the Exynos 2600 Without Fanfare

Only yesterday, specs of the new Samsung Exynos 2600 SoC leaked and now the company has unveiled the chip on its website. However, it has done so without any of the usual PR that goes with a new major product release like this. The leaked details were mostly correct and the Exynos 2600 will indeed feature a 3.8 GHz C1-Ultra prime core, three C1-Prox3 3.25 GHz performance cores and six power efficient C1-Prox6 2.75 GHz cores, all based on the Arm's latest v9.3 architecture. The Exynos 2600 will be fabricated using Samsung Foundry's 2 nm GAA process and its low power design and the package will use something Samsung calls heat path block which is using a High-k EMC material which is said to improve heat dissipation.The Exynos 2600 should deliver 39 percent better CPU performance compared to its predecessor according to Samsung.

The GPU will be Samsung's Xclipse 960—which yesterday's leaks also got correct—with support for ray tracing and Exynos Neural Super Sampling (ENSS) or AI-powered frame generation and upscaling. Samsung claims twice the compute performance of its predecessor here. There will also be a separate AI engine with a 32K MAC NPU and Post-Quantum Cryptography (PQC). Other features include support for LPDDR5X memory and UFS 4.1 storage, support for up to 4K or WQUXGA displays with a refresh rate of up to 120 Hz, native 8K 30p video encoding and 8K 60p video decoding and finally support for up to 320 MP single camera resolution and the ability to shoot pictures at 108 MP at 30 FPS and of course there are some AI features here too, that Samsung calls DVNR or Deep learning Video Noise Reduction. All modern video codecs such as AV1, VP9 and Samsung's own APV are supported, but there's no sign of H.266. We'll likely get more details of the Exynos 2600 at CES early next year.

Leaker Believes Samsung Exynos 2600 Mobile Chip Will Feature AMD "JUNO" iGPU

Earlier in the month, Samsung's marketing team initiated their Exynos 2600 promotional campaign. At the time, official teaser material did not provide any substantial hints about the next-gen mobile chipset's technical underpinnings. Weeks later, a fresh leak has laid out the cutting-edge mobile processor's fundamentals. PhoneArt's social media post seems to confirm previous predictions about an overall 10-core design, and being based on an important in-house 2 nm node process. Today's prediction did not outline the Exynos 2600 SoC's cluster configuration, but previous outpourings have settled on a "1x Prime + 3x Performance + 6x Efficiency" internal setup. PhoneArt reckons that the prime core will reach a maximum clock frequency of 3.9 GHz. A commenter, Erencan Yılmaz, reckons that this figure should be reduced to 3.8 GHz, due to power consumption considerations when looking at a 2 nm GAA-based design.

Supposedly, max. performance core speed is 3.25 GHz, while efficiency-oriented units can only reach up to 2.75 GHz. In the past Samsung and AMD have co-developed various Xclipse integrated graphics solutions, for proprietary smartphone platforms. Industry observers expected this RDNA-based collaboration to continue into a new generation. The fresh mentioning of a mysterious "AMD JUNO" iGPU has caused some confusion—this new Team Red IP is said to be clocked at 985 MHz. Additionally, PhoneArt claims that "JUNO" supports modern APIs: OpenGL ES 3.2, OpenCL 3.0, and Vulkan 1.3. It is plausible that "JUNO" could be a codename for the speculated "Xclipse 960" integrated graphics processor.

Rapidus Unveils New AI Design Tools for Advanced Semiconductor Manufacturing

Rapidus Corporation today announced a new suite of AI design tools to support its Rapidus AI-Assisted Design Solution (Raads), a key initiative to helping to realize its Rapid and Unified Manufacturing Service (RUMS) concept. The technology approach will also be rebranded to Rapidus AI-Agentic Design Solution with multiple tools being released starting in 2026 and will be provided to customers together with a process design kit and reference flows. The new tools include:
  • Raads Generator: An EDA tool based on large-scale language models (LLMs). When designers input semiconductor specifications, Raads Generator outputs register-transfer level (RTL) design data optimized for Rapidus' 2 nm manufacturing process.
  • Raads Predictor: A tool for RTL debugging and optimization for physical design, placement and routing. Raads Predictor will provide a power performance area (PPA) estimate in a short period of time.
Designers can use Raads Generator to write design ideas and desired specifications as RTL source code and then input data into Raads Predictor along with Synopsys Design Constraints (SDC), allowing them to predict the PPA of the silicon manufactured by Rapidus.

Marvell Designs Ultra-Low-Power 2 nm Dense SRAM, Outperforming Industry Standard

Marvell recently showcased its custom silicon IP advancements at the Marvell Analyst Day 2025. The company has developed SRAM IP that surpasses industry standards in terms of power efficiency and density. Initially launched in June, their 2 nm SRAM IP now reveals performance figures that highlight its superiority over standard solutions. In a 256K instance comparison, Marvell reports an 80% reduction in total power consumption, a 37% smaller area, and cycle times that are 22% faster. Additionally, Marvell's memory layout is more rectangular, facilitating easier integration into dense SoCs.

Further comparisons with top alternatives show that Marvell's custom SRAM uses 50% less area at the same bandwidth, reduces standby power by 66%, and delivers 17 times more bandwidth per mm² when normalized by area. These improvements are attributed to redesigned clocking and port structures that enhance bandwidth from on-die SRAM. Marvell argues that this architectural approach results in significantly higher bandwidth density and lower power consumption compared to standard dense SRAM IP. In an era where logic scaling continues to outpace memory in modern semiconductor nodes, having custom IP that aggressively boosts SRAM density and reduces power usage is a massive advantage.

AMD and Samsung Said to Discuss 2 nm Foundry Deal for Future EPYC CPUs

According to a report from Sedaily, cited by TrendForce, Samsung Electronics' Device Solutions (DS) Division is in discussions with AMD over manufacturing its chips on Samsung Foundry's second-generation 2 nm process known as SF2P. Samsung's 2 nm technology is competing directly with TSMC's N2 and Intel's 18A, both using the Gate-All-Around (GAA) transistor architectures. Industry sources say a decision on moving forward with a formal agreement could come as early as January next year, with actual production being more a matter of "when" rather than "if". As part of the talks, Samsung is expected to run AMD designs through a multi-project wafer (MPW) program in the near term. This would allow both companies to evaluate SF2P's performance and yields before committing to volume manufacturing. The chip involved is believed to be AMD's next-generation EPYC "Venice" server CPU, according to Global Economic News. If the MPW results meet AMD's expectations, sources say it could open the door for AMD to adopt a dual-foundry strategy, pairing Samsung with TSMC. Such a move would not be limited to server products, and could eventually extend to future consumer CPUs, including the "Olympic Ridge" Ryzen lineup.

Separately, the report highlights the AI-focused partnership between Samsung and AMD. Despite Samsung's challenges in entering NVIDIA's HBM supply chain, it has secured a strong position with AMD. Samsung is already supplying HBM3E 12-layer memory for AMD's MI350 accelerators and is considered well positioned for HBM4, which is expected to debut alongside AMD next-generation MI450 products. From Samsung's perspective, adding AMD as a foundry customer would further support its recent recovery. Sedaily notes that Samsung Foundry has picked up momentum after winning orders from major clients such as Tesla and Apple. At the same time, industry sources point to TSMC capacity constraints and rising wafer prices, those factors making Samsung an alternative.

Samsung's Proprietary "Heat Path Block" Tech Could be Adopted by Apple & Qualcomm

Samsung Foundry's "Heat Path Block" (HPB) technology is reportedly a key aspect featured within the upcoming Exynos 2600 application processor (AP). The South Korean's previous flagship mobile chipsets have often been beset with thermal issues, leading to the deployment of rival chipsets within certain generations of Samsung Galaxy smartphones. The South Korean seems to be pinning its hopes on the emerging Exynos 2600 across a number of fronts. Most notably as a confident "proof of concept" demonstration of the foundry business's cutting-edge 2 nm GAA process (aka SF2). According to a freshly-published ET News SK article, Samsung's innovative "semiconductor heat-suppressing technology" has already attracted interest from Apple and Qualcomm.

In the recent past, these VVIP clients have been on the order books for TSMC's best modern node processes. Samsung's SF2 line is seemingly being dangled as a carrot; designed to entice the return of former customers. Semiconductor industry trackers reckon that the Exynos 2600 SoC—enhanced with HPB tech—will serve as a strong demo of Samsung's 2 nm chip making capabilities. The foundry's in-house Heat Path Block is a copper heatsink that is packaged on top of a mobile AP. Heat dissipation is further maximized by moving DRAM to the side—previous Exynos layouts had this aspect placed atop APs. Insiders believe that a 30% improvement has been observed when comparing the Exynos 2600 to a predecessor. Samsung's System LSI division supposedly co-created the HPB design with an internal packaging development organization.

Samsung Teases Upcoming Exynos 2600 5G Mobile SoC, Insiders Allude to Regional Exclusivity

Samsung's next-gen flagship Exynos 2600 5G mobile chipset seems to be closing in on a launch window. Yesterday, a teaser trailer appeared online—the 30-second-long presentation mentioned that the South Korean giant's top SoC offering has been "refined at the core" and "optimized at every level." Contrary to this present day positive messaging, a steady flow of inside track information has indicated a problematic development. The Exynos 2600 processor's gestation has been linked to a troubled 2 nm GAA process (aka SF2), going back over a year. Late August reportage signalled that Samsung's foundry business was "going all in" with its cutting-edge 2 nm production line—future success could hinge on further node refinements; allegedly dubbed "SF2P" and "SF2P+." The Samsung Galaxy S26 series could be the first set of smartphones to land at retail with 2 nm silicon onboard. Previous Galaxy S generations—three in total—have relied entirely on Qualcomm chips.

Older in-house chip-powered efforts were viewed as inferior to rival flagship devices; normally operating with the best Snapdragon chipsets inside. According to a fresh Korea Joongang Daily article, a "dual-chip strategy" has been arranged for the forthcoming S26 models. The report alludes to yet another reliance on Qualcomm, although reduced for specific territories: "it is widely suspected by industry observers that the S26 and S26 Plus will be mounted with the Exynos 2600 in Europe and Korea, while the Snapdragon 8 Gen 5 Elite (3 nm) will be installed in the same models across the United States, China and Japan. For the S26 Ultra, Qualcomm chips will be used across all markets." Samsung's Exynos 2600 chip is expected to feature a 10-core design (1+3+6), paired up with a brand-new Xclipse 960 iGPU. This information was gleaned from early benchmark results; alleged evaluation sample data was discovered during the summer months.

Samsung HBM4 Development Complete, Mass Production Pending NVIDIA Approval

Samsung has reportedly finished development of its first sixth-generation HBM (HBM4) chips and is now in the final pre-mass-production stage. According to new reporting from AjuNews and Korean industry sources, Samsung's Device Solutions division has cleared Production Readiness Approval (PRA), the last internal checkpoint before full-volume manufacturing. HBM4 is expected to deliver roughly 60% higher bandwidth than the current HBM3E chips. Samsung has already shipped HBM4 samples to NVIDIA, which is evaluating them for use in its upcoming Rubin platform. Initial samples have already exceeded NVIDIA's next-gen GPU requirement of 11 Gbps per pin. The company is using improved 1c-class DRAM paired with a 4 nm logic base die, a combination that helps close the gap with rivals while keeping thermals and power in check at these higher speeds. Once NVIDIA signs off, Samsung can move straight into volume production, with manufacturing lines already prepared.

Samsung had already signaled strong progress during its Q3 2025 earnings call on October 30, noting that HBM4 samples were in the hands of global customers and that mass production is planned for 2026. "HBM3E is currently in mass production and being sold to all related customers, while HBM4 samples are simultaneously being shipped to key clients," the company said at the time. Samsung also confirmed that its foundry arm will prioritize stable 2 nm GAA output and HBM4 base-die production in 2026, alongside the ramp of its new Taylor, Texas fab. The company is also said to be working on a faster HBM4 variant targeting another 40% performance uplift, with an announcement possible as early as mid-February 2026.

NVIDIA Reportedly the Only TSMC A16 Node Customer

NVIDIA is reportedly the only customer queued for TSMC's next-generation A16 process, and the company is lining the node up for its upcoming "Feynman" GPUs. The company is lining up for samples in 2026 with volume ramps following in 2027, timing that would slot "Feynman" after "Rubin"-class products built on refined 3 nm variants. If the reports are accurate, NVIDIA's commitment would be notable as it would be the only major company opting for the "stopgap" node between N2 and A14. Customers like Apple are reserving TSMC N2 2 nm capacity and plan to transition to A14 as soon as it becomes available, bypassing the A16 node.

TSMC's A16 node is a nanosheet-focused design featuring enhanced backside power delivery, known as Super Power Rail (SPR). This design separates power routing from signal layers, reducing delivery losses. The process is expected to provide modest single-digit performance improvements, more noticeable power reductions for AI workloads, and slightly higher transistor density compared to the previous generation. For very large dies and high-power cards, these enhancements can lead to cleaner floorplanning, easier thermal management, and increased capacity for memory and interconnect bandwidth. While these improvements are significant for NVIDIA's datacenter chips, Apple's mobile and laptop chips may benefit even more from the less expensive N2 node and its variations.

Former TSMC Executive Reportedly Joins Intel with Sub-2nm Process Data

Former TSMC Senior VP Wei-Jen Lo has reportedly joined Intel as VP of R&D, according to a Liberty Times report cited by TrendForce. Lo is said to have started in late October, a move that has surprised many in the semiconductor industry. The report also claims that before retiring in July, Lo asked subordinates to brief him on and copy internal documents covering TSMC's 2 nm, A16, A14, and other next-generation process technologies. At the time, the requests appeared routine for someone at his level, but TSMC is now investigating and considering possible action. Local media say the case may draw scrutiny under national-security laws if the allegations hold up. If confirmed, this would be TSMC's second major 2 nm leak in recent months. On August 27, Taiwan's High Prosecutors Office indicted three TSMC engineers for stealing core technology files tied to the company's 2 nm development, seeking 7-14 years in prison. The concerns around Lo's move come as Intel pushes its 18A node into mass production this year and targets 14A production in 2027. TSMC, meanwhile, plans to begin mass production of A16 in the second half of 2026.

Lo worked at TSMC for 21 years and played a key role in building its technology portfolio. Commercial Times notes that the team accumulated more than 1,500 patents during his tenure, including about 1,000 in the U.S. He also had the trust of founder Morris Chang and stayed in his role until age 75, well past TSMC's typical retirement age of 67. Liberty Times reports that Lo's position at Intel involves overseeing advanced equipment and module development from R&D through pre-production, a role closely tied to yield improvement. The report also raises questions about whether Lo was bound by TSMC's usual 18-month non-compete agreement, with some sources suggesting he may not have signed one.

Update Nov 29th: Taiwan investigators raided the homes of former TSMC executive Wei-Jen Lo on Wednesday, seizing computers, USB drives, and other evidence under a search warrant, prosecutors said. On Thursday, a court approved the seizure of his shares and real estate. Today, Taiwan authorities have now frozen over NT$2 billion (US$64 million) worth of Lo's assets, including NT$1.8 billion in TSMC stock, China Times reports. Intel continues to deny TSMC's allegations and said it fully supports its new hire.

Speculative Snapdragon 8 Elite Gen 6 iGPU Looks Promising, Despite "sub-Pro" Performance

Qualcomm introduced its Snapdragon 8 Elite Gen 5 mobile chipset at the tail end of September, followed by yesterday's presenting of their Snapdragon 8 Gen 5 platform. On a surface level, this confusing naming scheme makes it difficult to tell the difference between the two fifth-gen options. Complicating matters further, semiconductor industry soothsayers have already moved onto predicting the nature of next generation Qualcomm offerings. This week, Digital Chat Station reckoned that Qualcomm's engineering department has moved onto the development of a mysterious "Snapdragon 8950" SoC. This forthcoming "standard" SKU could end up becoming the "Snapdragon 8 Elite Gen 6" chip; set to exist alongside a more potent "Pro" model.

Earlier today, Digital Chat Station shared fresh musings via their Weibo blog: "the Snapdragon 8 Gen 5 platform's performance was pretty good, which gives me some confidence in next year's standard Snapdragon 8950. After all, it uses the same high-end TSMC 2 nm '2+3+3' Oryon CPU (as the Gen 6 Pro), with a slightly downgraded GPU. This chip will be positioned directly in the 4000 yuan (~$560 USD) price range." The speculated "Snapdragon 8 Elite Gen 6 Pro" option is expected to utilize a more powerful Adreno GPU, thus demanding a price premium. Insiders believe that Qualcomm will be selecting TSMC's newer 2 nm "N2P" for production of the sixth-gen Snapdragon 8 production. Reportedly, MediaTek leadership has made a similar decision regarding the makeup of its next flagship designs.

Rapidus Targets 2029 Production of 1.4 nm Chips, Plant Construction to Begin in 2027

Rapidus, Japan's most ambitious chipmaker, has reportedly planned to begin constructing a second factory in Hokkaido in 2027, with the production of its most advanced 1.4 nm chips expected to start in early 2029. According to Nikkei, this project is a central part of a trillion-yen initiative supported by substantial government funding. The new facility will follow the company's first plant in Chitose, which is on schedule to begin mass production of 2 nm chips by late 2027. Rapidus reportedly wants to commence intensive research on 1.4 nm technology in 2026, continuing its collaboration with IBM, which has provided the technological foundation for its advancements.

Update 16:00 UTC: A Rapidus spokesperson reached out to us with the statement below:
Rapidus SpokespersonAlthough there have been recent media reports regarding construction and operation of a 1.4 nm-node semiconductor factory from Rapidus, these articles are speculation and did not originate from our company. We do not comment on rumors or speculation. Any updates on our roadmap will come directly from our company and if we decide on any development that should be made public, we will announce it promptly.

TSMC to Build Three New 2 nm Fabs in Taiwan for $28 Billion

Liberty Times reports that TSMC is preparing to build three more 2 nm fabs in Taiwan, with a combined investment of roughly NT$900 billion (about $28 billion). The sites will likely be located near the Southern Taiwan Science Park. According to the report, TSMC originally planned seven 2 nm fabs, two in Hsinchu's Baoshan district and five in Kaohsiung's Nanzi district. The three new facilities would raise the total to 10, as demand for advanced nodes continues to climb. Groundbreaking could start as early as next year, pending land acquisition and environmental approvals. TSMC began initial 2 nm production this quarter and is already expanding its next nodes. The company broke ground on its first 1.4 nm fab in Taichung in November with volume production aimed for the second half of 2028. The A14 (1.4 nm) node is TSMC's next major process milestone after the 2 nm node and is expected to deliver up to 15% higher performance at the same power. Alternatively, at similar speeds to the 2 nm process, the 1.4 nm node will use 30% less power.

The company is also accelerating its U.S. plans. During its October earnings call, Chairman C.C. Wei said TSMC will bring N2 and more advanced technology to Arizona faster than originally scheduled. Nikkei recently reported that TSMC is close to securing more land next to its latest Arizona fab, signaling further expansion beyond its current $65 billion investment. As previously reported, TSMC expects the third U.S. fab to start 2 nm and A16 production in 2027, one year earlier than planned. The TSMC push on its U.S. expansion strategy is likely a result of competition from Intel's 18A and the upcoming 14A, however, U.S. fabs will remain one node generation behind the Taiwanese ones.

Tachyum Intros 2 nm Prodigy Universal Procesor Targeting Improved AI Rack Efficiency

Tachyum today announced details and specifications for its 2 nm Prodigy Universal Processor, which will enable AI models with parameters many orders of magnitude larger than those of any existing solution at a fraction of the cost.

Prodigy Ultimate provides up to 21.3x higher AI rack performance than NVIDIA Rubin Ultra NVL576. Prodigy Premium provides up to 25.8x higher AI rack performance than Vera Rubin 144. Technical details of the 2 nm Prodigy, the first ever chip to exceed 1,000 PFLOPs on inference, will be published within a week. NVIDIA Rubin delivers 50 PFLOPs.

Japan Ramps Up Photoresist and MOR Capacity for 2 nm EUV Lithography

As the 2 nm-class nodes are starting to enter volume production, Japanese material suppliers are expanding the capacity of EUV-capable resists and high-purity process chemicals, according to Nikkei. Tokyo Ohka Kogyo (TOK), Adeka, and JSR are building or upgrading fabs for photoresists and metal-oxide resist (MOR) chemistries that improve resolution, line-edge roughness (LER), and etch resistance for extreme ultraviolet patterning. TOK will invest ¥20 billion ($130 million) to add a South Korea photoresist plant (operations targeted for 2030), expanding capacity up to three to four times for packaging and memory customers such as Samsung, TSMC and SK hynix, and a separate ¥12 billion ($78 million) site for high-purity semiconductor chemicals. These chemicals (solvents, developers, surfactants, cleaning agents) are critical for EUV processes where resist sensitivity, contamination control, and outgassing limits directly affect wafer yield and mask lifetime. Adeka's ¥3.2 billion ($20 million) investment will equip its Ibaraki lines for MOR mass production (from ~April 2028), while JSR's MOR facility in South Korea is slated for service by late 2026.

These expansions address bottlenecks unique to EUV lithography technology, such as resist sensitivity vs resolution tradeoffs, defect mitigation and contaminant control. Japan's suppliers already account for the majority of high-end photoresist supply (~91% market share), giving them an edge over competitors that have improved i-line/KrF resist technology but still lag on EUV MOR and process-grade chemical scale-up. MORs contain metal compounds that increase EUV photon absorption and improve pattern transfer: higher absorption reduces dose requirements, and metal-based chemistries can yield better LER and etch durability vs conventional organic resists, a key for sub-2 nm pattern fidelity.

TSMC to Begin Construction of Its Next-Gen A14 (1.4 nm) Fab in Taiwan

TSMC is set to begin construction of its upcoming A14 fab plant in Taichung, Taiwan, as the company has received all necessary permits, Focus Taiwan reports. The project will include the main fab, a central utility plant for power and water management, and several support buildings. The A14 (1.4 nm) node is TSMC's next major process milestone after the 2 nm node and is expected to deliver up to 15% higher performance at the same power. Alternatively, at similar speeds to the 2 nm process, the 1.4 nm node will use 30% less power. The A14 node is designed mainly for AI acceleration and high-performance computing chips, with TSMC targeting mass production in the second half of 2028. The new fab is estimated to be a $49 billion single-site investment from TSMC, one of its largest endeavors.

TSMC recently confirmed that its most advanced N2 (2 nm) node will begin volume production by the end of the year, with a rapid increase planned through 2026 in Taiwan. The announcement follows record Q3 revenue of $33.1 billion, up 41% year-over-year, results mainly driven by strong demand for advanced nodes. Current 3 nm (N3), 5 nm (N5), and 7 nm (N7) nodes remain popular due to increased market demand, with N3 shipments accounting for 23% of wafer revenue, N5 for 37%, and N7 for 14%.
Return to Keyword Browsing
Nov 22nd, 2026 17:51 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts