Market reports suggest that MediaTek's tensor processing unit (TPU) v9 project with Google could face delays in volume production, reportedly because Google is revising the chip's design; the title also cites glass substrate yield issues. According to the title, 2026 orders reportedly remain strong. This is an unconfirmed market report/rumor, and the volume production delay has not been verified. Industry observers expect MediaTek may address some of these concerns during its earnings conference in late October.
Security researcher Quentin Kaiser disclosed two bugs in Broadcom's proprietary Wi-Fi implementation, dubbed "WPSplinter," dating back to at least 2010. The bugs lie in Broadcom's handling of Wi-Fi Protected Setup (WPS): specially crafted WPS messages can corrupt the router's memory, allowing an attacker within Wi-Fi range to run their own code on the device without needing the password. According to Kaiser, Broadcom knew about the vulnerability in 2018 and released a partial fix without public disclosure, subsequently recommending manufacturers switch to the open-source hostapd software. Kaiser demonstrated working exploits against routers supplied by Belgian providers Proximus and Orange — on an ASKEY TCG3000 gateway he ran code and extracted the Wi-Fi password — and also found the flaw in Netgear and Technicolor routers supplied by VOO. He estimates around 850,000 Proximus gateways, 500,000 at VOO and 210,000 at Orange Belgium could be affected, roughly 1.56 million routers in Belgium alone (rough estimates, he admits), with worldwide numbers potentially considerably higher. Routers running older eCos-based chips (BCM3380, BCM3383, BCM33834) may never get a fix — Kaiser calls the flaw a "forever day" — while routers using Linux-based BCM3390, BCM65xx, and BCM68xx chips are also at risk unless manufacturers switch to hostapd. Kaiser first reported the vulnerabilities to Broadcom on July 18 with a second report on July 27 and a 90-day disclosure window; Broadcom responded on July 28 saying a patch was available, then declined to assign a CVE identifier; CERT/CC closed its case on August 4. By October 15, the day before his presentation, Kaiser said none of the affected devices he had identified had received a patch. He has not reported evidence of attacks in the wild. The source does not state any price or supply impact.
Google’s TPU v9 involving MediaTek could face delays, with shipments potentially slipping to 1Q28; meanwhile, ByteDance may purchase data center chips from MediaTek. The source does not state the cause.
Anthropic proposes offering Blackstone more than 10% total return to finance a US$60 billion artificial intelligence (AI) chip procurement package, as higher infrastructure financing costs make expanding computing capacity more expensive. The proposed arrangement, reportedly code-named Project North Star, would fund purchases of Google tensor processing units (TPUs) with a combined capacity of approximately 1.5 GW.
Google tensor processing units (TPUs)GoogleDigiTimes AsiaPartnership/M&A
According to a Wccftech interview with NVIDIA's Vera team reported on 2026-10-19, NVIDIA has chosen a monolithic compute die for its Vera CPU over the chiplet architectures used by major x86 competitors; Vera also features a custom LPDDR5X memory subsystem and NVIDIA's second-generation Scalable Coherent Fabric. As the report explains, unlike traditional cloud workloads that prioritize high core counts and throughput, agentic AI requires strong single-threaded performance, high memory bandwidth, and low latency to execute sequential tasks quickly and keep GPUs highly utilized. The monolithic design allows CPU cores to communicate without crossing boundaries between separate chiplets, reducing data movement latency. NVIDIA says Vera delivers 3.4 TB/s of bisection bandwidth, three times that of the leading x86 processor used in its comparison; Vera delivers 1.5 times the performance of its predecessor, Grace; and for agentic workloads involving sequential tasks such as Python scripting and code compilation, the company claims up to twice the performance of leading x86 processors, with a 1.6-times advantage on a geometric-mean basis. Vera also introduces NVIDIA's Spatial Multithreading technology. NVIDIA initially positioned its Grace CPU primarily as a host processor for GPU systems, but the report states that growing demand for agentic AI and reinforcement learning has prompted a shift toward standalone Vera CPU racks. Separately, according to the Central News Agency, NVIDIA Senior Vice President of Personal Computing Jeff Fisher denied during an interview in Taiwan rumors that NVIDIA plans to discontinue its flagship GeForce RTX 5090 graphics card, calling them untrue and stressing that the RTX 5090 remains a key product that NVIDIA will continue to supply; he acknowledged supply constraints are affecting the global industry, not just NVIDIA, but said the company is doing everything possible to bring products to market. TechPowerUp adds that the RTX 5090 remains difficult to find near its US$1,999 MSRP, with some listings on Amazon and Newegg exceeding US$8,000.
According to Commercial Times, citing institutional investors, Google's TPU v9 project, involving MediaTek, could see its initial tape-out delayed until January 2027 following design revisions, with shipments potentially pushed back to the first quarter of 2028; the November 2026 sampling target remains unchanged, and the production-version tape-out is expected in the second quarter. The report notes that revisions to Google's TPU v9 Core A design, codenamed Humifish... wait, Humufish, have delayed development by approximately three weeks, affecting the Core B and I/O designs handled by MediaTek. According to TrendForce, the chip is expected to adopt Intel's EMIB-T advanced packaging, with substrate yields remaining a key challenge to production scalability and cost competitiveness. Bloomberg, citing sources, reports that ByteDance is nearing an agreement to purchase MediaTek's data center chips, potentially making it MediaTek's second major ASIC customer after Alphabet's Google. Among Taiwanese ASIC design firms: GUC, supported by cloud customers' CPU and ASIC programs, reported record September revenue of NT$7.145 billion, up 97.8% year over year; Alchip posted September revenue of NT$9.979 billion, surging 496% year over year as shipments of 3nm AI accelerators increased. Faraday has obtained Samsung's 2nm process design kit (PDK) to pursue related design opportunities; per Commercial Times, Samsung's 4nm capacity is reportedly fully utilized, prompting some AI ASIC customers to evaluate its 2nm node. Faraday has also completed its design environment setup for Intel's 18A-P and begun offering design services, with plans to pursue more projects in 2027. Commercial Times, citing institutional investors, projects global AI ASIC accelerator shipments to reach 14.1 million units in 2027, up approximately 69% year over year.
Google TPU v9Core A (codenamed Humufish)Core BMediaTek, Google and 11 totalTrendForce
According to Tom's Hardware, published 18 October 2026, a project-by-project audit shared by SemiAnalysis on 17 October 2026 finds that more than 300 local US data center construction moratoriums have had limited impact on planned buildouts: only about 2.3 GW of the roughly 21 GW of data centers in moratorium areas (about three projects) is likely to be delayed, and this figure excludes the Texas Commission on Environmental Quality (TCEQ) pause. The firm forecasts a total of 38 GW of US data center IT capacity coming online in 2027, more than double that of 2026; of that, 22 GW is under vertical construction and the other 16 GW is mostly financed with site work being done. The enacted local moratoriums cover about 20 GW, with only about 1.5 GW of it delayed; New York's pause adds 0.8 GW more. As of September, the firm tracked more than 400 local instruments across 17 states, the enacted ones including a few permanent zoning bans; this summer New York, Texas, Pennsylvania, and Oregon acted within under two months. Reasons given for the limited effect include that moratoriums do not revoke existing approvals, short ones may expire before later projects need approval, litigation or equipment lead times may already set the schedule, and capacity already under construction or fully permitted is out of reach. Texas Gov. Greg Abbott told the TCEQ to issue no permits for data center projects, and law firm Bracewell expects the pause to last until ERCOT finishes its review in mid-December; the TCEQ's compliance report to Abbott's office is due Monday.
As reported by Tom's Hardware on 17 October 2026, after rumors that RTX 5090 production was stopping, Nvidia senior VP Jeff Fisher reportedly said in a purported post-event interview that the card has not been discontinued and will continue to ship. He reportedly stated: "The supply chain (capacity) is indeed somewhat constrained, but NVIDIA is doing everything it can to get the product on the market... production has not been discontinued, and NVIDIA will continue to ship," while noting other variants of the RTX 5090 (chip) exist. Earlier rumors held that, after Nvidia reportedly put the RTX 50 Super series on hold, production of the already-rare RTX 5090 was stopping, likely to use precious silicon manufacturing capacity for more AI accelerators. Retail spot-checks: RTX 5090s on Amazon sell for about $8,500 from third-party sellers; Newegg lists 5090s for about $7,300, likely from Chinese sellers judging by shipping time, with U.S. tariffs likely adding to the price; Best Buy has zero cards on offer; and as of this writing the card remains absent from Nvidia's own marketplace and Micro Center. Tom's Hardware received no reply from Nvidia for comment; the claims are reported second-hand, and production status is not officially confirmed.
On October 17, 2026, Taiwan announced it will add AI servers and the latest chips to its controlled export list to align its policies with U.S. trade bans; the revised Foreign Trade Act will impose penalties of up to five years imprisonment and/or a fine of about $92,500 for smuggling these items once the controls take effect. The move follows a U.S. tariff agreement and a raid on Super Micro, after U.S. authorities accused Super Micro employees of smuggling $2.5 billion worth of AI servers; months later Taiwanese authorities raided Super Micro’s Taipei office, leading to the indictment of nine individuals. Taiwan currently does not treat AI chip exports to China as a crime, and the new controls are not yet in effect.
According to Tom's Hardware, published 16 October 2026, Chinese startup Langdao Silex (established in 2024 by Tsinghua University's Gas Discharge Plasma Laboratory alumni) demonstrated a prototype EUV light source with over 15W of output power, 90% domestic components, and decent uptime at the 26th China International Industry Fair in Shanghai, without lighting it up. The FlintNano LPP-OXXO system uses laser-produced plasma (LPP) technology with a 1,000W driving laser operating at 1kHz–20kHz, conversion efficiency exceeding 2%, brightness of 100 W/(mm²·sr), continuous operation of over 100 hours, plasma stability of 3% RMS and EUV output power stability of 5%; however, its output remains insufficient for high-volume chip production, and the company has not disclosed whether the 15W is measured at the plasma or at intermediate focus. Per TrendForce, citing various sources, the company will need another three to five years to commercialize its technology. Langdao is also developing the higher-power FlintMicro-LPP EUV source, which has completed proof-of-principle validation, but its output power, conversion efficiency and operating lifetime are undisclosed. A light source alone does not make an EUV scanner; a complete EUV lithography scanner would require further advances in collection/illumination optics, multilayer mirrors, wafer and reticle stages, and system integration.
According to a deal post published by Tom's Hardware on 16 October 2026, the Corsair Vengeance RGB DDR5 32GB (2 x 16GB) memory kit (DDR5-5600, CAS latency CL40, with RGB lighting and Intel XMP profiles) is now priced at $409.99, down from $517.99 — a $108 cut, or 21% off — available direct from Corsair. The author verified via Tom's Hardware's own price tracking and other online resources that this is the cheapest available DDR5 5600 memory kit currently purchasable. The article notes as background that RAM and storage prices had previously risen sharply amid the ongoing AI boom, and states the author's view that prices may rise further and that production may not normalize for up to a decade; these are contextual assertions, not confirmed facts. The specific cause of this price cut is not stated by the source (the source does not state the cause).
Corsair Vengeance RGB DDR5 32GB (2 x 16GB) 5600 memory kitCorsairTom's HardwarePrice Movement
Newegg is discounting an all-AMD 4K gaming PC from CyberPowerPC: the title states a $260 cut, and the article says the rig is now $2,299; a separate deal box shows the CyberPowerPC GM70929 Gaming PC reduced from $2,549.99 to $2,259.49, saving $290.50 (11%). The build includes an AMD Ryzen 7 9800X3D, AMD Radeon RX 9070 XT, 32GB of DDR5 RAM, and a 2TB Gen 4 NVMe SSD, plus a 1000W power supply, ARGB liquid cooling and a B850 motherboard, with a free copy of Onimusha listed at $69.99 at Newegg. The article notes that RAM and SSD price hikes and shortages continue to bite PC builders and that this PC's price is still higher than a few months ago; it also states the 9800X3D was recently dethroned as the fastest gaming CPU by the 9850X3D, that the RX 9070 XT is currently worth $800 in the market, that the 32GB DDR5 RAM is worth at least $400-$450, and that the 2TB SSD is worth $250. The source does not state the cause of the discount.
According to Reuters, SK hynix is reportedly asking the U.S. government to authorize additional semiconductor manufacturing equipment shipments to its Dalian facility in China in exchange for a commitment to build a new NAND fab in the U.S. The Dalian fab is the world's only facility to produce 192-layer floating gate 3D NAND for Solidigm's data center-grade SSDs for AI, enterprise, and hyperscale data centers; its manufacturing operations rely on equipment from American suppliers including Applied Materials, KLA, and Lam Research, whose exports to China require a U.S. government license. Solidigm has requested U.S. export licenses for additional tools, but the U.S. government has yet to authorize the shipments; last December the U.S. Commerce Department granted SK hynix a limited authorization for equipment deliveries to Dalian for 2026, but has not issued a similar equipment export authorization for 2027. Solidigm wants to install additional manufacturing equipment in the unused portions of its Dalian Mod B facility, completed in 2023, and is also considering an initial public offering as early as 2027 to raise additional capital. The proposed U.S. fab and the equipment export arrangement remain reported proposals, not approved or implemented actions.
3D NAND192-layer floating gate 3D NANDdata center-grade SSDsSK hynix, Solidigm and 5 totalTom's HardwarePolicy/Regulation
Related manufacturers
SK hynix, Solidigm, Applied Materials, KLA, Lam Research
According to ZDNet, SK hynix has unveiled its ultra-lightweight Hynix Transaction Processing Unit (H-TPU) architecture, dedicated to SSD data-plane control and transactions, distributing storage-control tasks across dozens of compact H-TPU clusters. According to Vice President Koo In-jae, who oversees SoC design, H-TPU eliminates complex CPU features such as superscalar and out-of-order execution and adopts a simplified four-stage pipeline, reducing processor area to approximately 4% and power consumption to around 2.5% of a conventional CPU core, activating only required modules. Building on H-TPU, SK hynix plans to introduce hardware automation (HWA), assigning repetitive storage-control tasks such as instruction processing, data address lookups, and temporary buffer copying to dedicated hardware circuits. SK hynix expects the H-TPU-based controller architecture to complement high-bandwidth flash (HBF), an emerging memory technology for AI workloads; AI inference is highly read-intensive with read-to-write ratios reaching approximately 300,000 to 1, and AI compilers and runtimes could inform the HBF SoC of upcoming data-processing needs for prefetching. Separately, according to Yonhap News, Samsung began mass production of its PCIe 6.0-based PM1763 enterprise SSD in July 2026, featuring a newly developed 4nm controller and ninth-generation V-NAND, delivering more than twice the performance and over 1.8 times the power efficiency of its predecessor.
Sipeed announced its MaixStation, a Chinese-made rival to Nvidia’s DGX Spark and AMD’s Ryzen AI Halo, expected to cost around $3,999 for the 128GB model. The company said sign-ups are open ahead of a Kickstarter campaign set for the end of 2026, with units to ship within three months after the campaign closes. The NPU toolchain is due by the end of October 2026. The announcement was published on 15 October 2026.
According to an Oct. 14 Bloomberg report cited by the article published 15 October 2026, the EU is in talks with China to lift restrictions China imposed on Sindlhauser Materials GmbH, a German precision-manufacturing and advanced-coatings firm in ASML's supply chain; in return, the EU would remove restrictions it placed on one unnamed Chinese firm accused of aiding Russia. The swap was discussed during EU Commissioner for Trade and Economic Security Maroš Šefčovič's visit to Beijing; the EU is expected to make the first move, with any delisting on its end requiring unanimous support from member states, and both sides agreed to share more information and give advance warning on future export restrictions. In July 2026, China put Sindlhauser, along with 13 other European companies, on its export-control list in apparent retaliation for the EU's 21st Russia sanctions package, which sanctioned several Chinese firms accused of supplying Moscow with components usable for weapons; Beijing's official reason was the firms' alleged dealings with Taiwan. The restrictions ban Chinese exporters from selling dual-use items to the firm and prohibit foreign companies and individuals from passing on China-origin dual-use items to it, blocking third-party sourcing. Sindlhauser relies almost entirely on Chinese rare earths and critical minerals; the German and Bavarian state governments have helped explore workarounds such as alternative suppliers and recycling existing materials, but the company risks going out of business in the coming months if the sanctions remain.
Chinese rare earthscritical mineralsdual-use itemsSindlhauser Materials GmbH, ASMLTom's HardwarePolicy/Regulation
ASML published its Q3 2026 earnings on October 14, 2026, saying the unprecedented spending and investment brought about by the AI boom is not slowing down, with customers already lining up to place orders for its advanced chipmaking tools for 2029 and 2030 output; CEO Christophe Fouquet said the company would struggle to meet demand despite plans announced in July to raise capacity by 30% in 2027 and another 30% in 2028. The company also confirmed it will raise prices starting in 2027, with the full effect showing in 2028. The Q3 2026 report detailed a net profit of €4.96 billion on net sales of €11.7 billion; the company projects Q4 sales of €14.2 billion to €15.2 billion ($15.9 billion to $17 billion) and boosted its full-year sales projections from at least €43 billion to €44-45 billion. ASML's shares rose 2% on Wednesday, up almost 70% so far this year, with its €586 billion market valuation making it Europe's largest firm by market cap. Additionally, Terafab will reportedly receive its first EUV chipmaking tool from ASML in mid-2027; CFO Roger Dassen said it is "still kicking the tires on the exact size of that (Terafab) opportunity."
Summary: TSMC CEO C.C. Wei confirmed on Thursday, October 15, 2026 that the company is considering expanding its American production capacity beyond Arizona, with Texas as one option; teams have been sent to Texas for evaluation, and TSMC has discussed with a customer attempting to build a fab called Terafab (Elon Musk). If TSMC joins the project, it would insist on operating the fab. Discussions are ongoing and no agreement has been reached yet.
On October 15, 2026, SK hynix is disaggregating SSD controller cores, developing H‑TPUs (Hynix‑Tensor Processing Units) to offload low‑level read/write command processing, metadata management, and buffer control from the main controller CPU, and plans to deploy H‑TPUs in its next‑generation enterprise SSDs and Storage Next SSD controllers.