Apple explores Chinese memory for iPhone 18/MacBooks, countering 100% price hikes (iPhone 17 LPDDR5X) & supply uncertainty post-Q1 2026. This aims to boost flexibility, strengthen negotiation, and could disrupt global markets, signalling competitive Chinese quality.

Apple is reportedly evaluating Chinese memory suppliers for potential use in the iPhone 18 series and MacBooks. This strategic move aims to enhance supply flexibility and strengthen its negotiating position against dominant suppliers like Samsung Electronics, SK hynix, and Micron Technology amid rapidly climbing memory prices and tightening supply visibility. Evidence includes Apple accepting a 100% price hike for LPDDR5X memory from Samsung for the iPhone 17 series. Some NAND quotes have reportedly doubled, and supply is secured only through H1 2026 (DRAM) and Q1 2026 (NAND), leaving future conditions uncertain. Negotiations have also shifted from semi-annual to quarterly cycles, indicating suppliers' firm stance. Currently, Samsung, SK hynix, and Micron primarily supply DRAM, while NAND mainly comes from Samsung, SK hynix, and Kioxia. Introducing Chinese suppliers could provide critical leverage. Beyond price, this shift carries significant global implications. With ~240 million iPhones sold annually, adopting Chinese memory could materially disrupt market share and signal competitive Chinese quality globally. While US-China tech tensions make near-term US deployment unlikely, Apple could use Chinese memory for devices sold within China. The industry closely watches if Apple gains pricing leverage, potentially prompting other major tech firms to diversify their supply chains.

Samsung is reportedly ending 2D NAND production at Hwaseong Line 12 (80K-100K wafers/month) by March, converting it to a 1C DRAM fab. This strategic shift, supported by its Xi'an plant's advanced 3D NAND transition and SK hynix's aggressive 1C DRAM investments (e.g., KRW 21.6T), signals a industry focus on high-performance DRAM.

Samsung is making a significant strategic pivot, reportedly halting its 2D NAND production at Hwaseong Line 12 as early as March. This facility, capable of 80,000–100,000 12-inch wafers monthly, will convert to a 1C DRAM fab to address bottlenecks. This move, communicated to customers last year, officially ends Samsung's 2D NAND chapter, which began in 2002. Its 3D V-NAND launched in 2013; now, ~13 years later, its last 2D NAND facility closes. The reduced NAND volume in South Korea will be offset by Samsung's Xi'an plant in China (~40% of total NAND output). This plant rapidly advances, transitioning from 128-layer (6th-gen) to 200-layer+ (8th-gen V8) 3D NAND processes. This strategic shift aligns with Samsung's broader focus on 1C DRAM, crucial for next-gen HBM4 and expanding into general server products. Samsung boosts 1C DRAM investments across Pyeongtaek (P4 converting to solely DRAM) and Hwaseong (older lines upgrading). Archrival SK hynix also aggressively ramps 1C DRAM, projecting 170,000–200,000 wafers/month, up from its initial 90,000 wafer/month plan. SK hynix's M16 fab will convert its 1a DRAM line to 1C DRAM by next year, with the 1b line following. SK hynix announced a KRW 21.6 trillion (USD 15 billion) investment by Dec 2030 in the Yongin Semiconductor Cluster, part of a total KRW 31 trillion (USD 21.5 billion) commitment.

NASA advanced Artemis II with a VAB rollback, unveiled Starliner investigation findings on Feb. 26, and received ASAP's 2025 report warning of workforce/acquisition challenges. Critically, Joel Montalbano was named acting SOMD lead, Oman became the 61st Artemis Accords signatory, and a $5M STEM award spurred future innovation.

NASA made significant announcements this February across human spaceflight, safety, and exploration. The agency will host a news conference on Friday, Feb. 27, at 10 a.m. EST from Kennedy Space Center to detail "Next Steps for the Artemis Campaign," following the Artemis II SLS rocket and Orion spacecraft's rollback to the Vehicle Assembly Building. Preparations for Artemis II also included coverage of a wet dress rehearsal on Feb. 19, 2026, and early results discussed on Feb. 3, 2026. Crucially, NASA released findings from the Boeing CST-100 Starliner Crewed Flight Test investigation on Thursday, Feb. 26, acknowledging challenges in its uncrewed and recent test flights. Also on Feb. 26, Joel Montalbano was appointed acting associate administrator for the Space Operations Mission Directorate (SOMD). Space safety remains paramount, as the Aerospace Safety Advisory Panel (ASAP) released its 2025 annual report. While recognizing safety achievements, ASAP warned of significant challenges stemming from "interconnected factors" including workforce, acquisition, infrastructure, and space support. In active missions, a SpaceX Dragon spacecraft departed the International Space Station (ISS) on Feb. 26 as part of the 33rd resupply mission, returning scientific research samples. Furthermore, NASA’s SpaceX Crew-12 mission launched on Feb. 20, 2026, at 5:15 a.m. EST for a science expedition to the ISS. Looking ahead, NASA selected Vast for the sixth private astronaut mission (summer 2027) and Axiom Space for the fifth (January 2027) to the ISS. Other key developments include NASA’s 15th consecutive ‘clean’ financial audit opinion for fiscal year 2025 and new Artemis payloads for lunar terrain, radiation, and history studies. NASA also honored fallen heroes on Jan. 22, 2026, Day of Remembrance and recognized the retirement of astronaut Suni Williams on Dec. 27, 2025, after 27 years and three ISS missions.

Smith+Nephew partners with Si-Bone for iFuse TORQ TNT, boosting its pelvic fracture fixation portfolio. This FDA Breakthrough Device, eligible for Medicare NTAP, strengthens S+N's trauma position, offering next-gen solutions for sacrum/ilium, improving patient access and outcomes.

Smith+Nephew recently announced a strategic distribution agreement with Si-Bone, specifically focusing on the iFuse TORQ portfolio. This collaboration significantly expands Smith+Nephew's portfolio for percutaneous pelvic fracture fixation procedures, including those involving the sacroiliac (SI) joint. This move solidifies the London-based orthopedic giant's position in high-frequency, high-impact trauma procedures. The iFuse TORQ portfolio, which includes the iFuse TORQ TNT and TORQ implants, is engineered for pelvic fracture fixation. The iFuse TORQ TNT system particularly addresses the anatomical and bone density needs of the sacrum and ilium, representing what Smith+Nephew calls the "next generation of technology" in this treatment area. A critical aspect of the iFuse TORQ TNT system is its FDA breakthrough device designation and eligibility for a Medicare new technology add-on payment (NTAP) for reimbursement. This helps facilitate broader access for patients needing these advanced solutions. Scott Gunn, Smith+Nephew VP, Trauma, Extremities and Shoulder, stated, "This collaboration reflects a shared commitment to innovation, clinical excellence, and improving patient outcomes. By combining SI-BONE’s industry leadership with Smith+Nephew’s established commercial platform, we see a meaningful opportunity to expand access to differentiated solutions, deepen surgeon engagement, and drive sustainable growth." This partnership enhances both companies' ability to serve surgeons and patients in a high-need area.

Taiwan's PSMC, Intel, and SoftBank are developing "Z-Angle Memory" (ZAM) to revolutionize AI/HPC. Aiming beyond HBM, ZAM promises 40-50% less power, 512GB/chip, with prototypes by 2027. This collaboration could break existing memory strongholds and establish a new industry roadmap.

The AI boom is fueling explosive demand for advanced memory, putting a spotlight on a new tri-nation alliance. Taiwan’s PSMC has partnered with Intel and SoftBank Group on a significant AI memory collaboration, aiming to challenge the current market stronghold of Samsung, SK hynix, and Micron. This Taiwan-US-Japan joint initiative focuses on "Z-Angle Memory" (ZAM) technology, specifically designed for AI and high-performance computing (HPC) applications. ZAM aims to overcome existing HBM limitations in capacity, power efficiency, and data transfer speeds. Early reports indicate ZAM's game-changing advantages: it could cut power consumption by 40-50%, streamline production with Z-Angle interconnects, and pack up to 512 GB per chip. Intel has noted that conventional memory architectures struggle to keep pace with AI's growing demands, emphasizing ZAM's reliance on multi-layer stacking and advanced packaging. The collaboration features a clear division of labor: SoftBank’s Saimemory handles design and IP, Intel contributes expertise in stacking and memory architecture (including its Next Generation DRAM Bonding initiative), while PSMC and Japan’s Shinko Electric Industries will manage pilot production and manufacturing. The partners target prototype development by 2027, with commercial mass production slated for 2029. ZAM made its first public appearance at Intel Connection Japan 2026 on February 3. This strategic push seeks to establish an alternative memory roadmap beyond HBM, reducing dependence on current supply chains and positioning PSMC as a pivotal player in future AI memory manufacturing.

TSMC is rapidly expanding with up to 10 new fabs across Taiwan, focusing on sub-2nm nodes (2nm, 1.4nm) and advanced packaging. A $52-56B capex for 2026 fuels this build-out, critical for future semiconductor supply and supporting industry leaders like NVIDIA.

TSMC is driving a colossal expansion with plans for up to 10 new fabs across Taiwan, focusing on cutting-edge nodes (2nm, A16, A14) and advanced packaging, critical for partners like NVIDIA ahead of its Q1 earnings and GTC 2026. Hsinchu's Fab 20 is central to sub-2nm development. P3/P4 civil engineering is underway. This 2nm hub, already in production since late 2025, is set to deliver 20,000–25,000 wafers monthly. Central Taiwan's Fab 25 is becoming a 1.4nm powerhouse with four planned fabs. Piling began late 2025; risk production is anticipated by late 2027; full volume by H2 2028. Advanced packaging fabs, AP5B (Taichung) and AP7 P1 (Chiayi), are also due for completion in 2026. Southern Taiwan is accelerating 2nm production. Kaohsiung's Fab 22 saw P1 enter volume production in H2 last year, P2 is in trial, and P3's structure is mostly complete. P4 and P5 are under construction, with all five fabs aiming for full operation by Q4 2027. Additionally, Tainan plans new 2nm fabs, with construction potentially beginning in May, pending an April environmental review, to meet surging demand. This domestic expansion is fueled by TSMC's guided US$52–56 billion capital expenditure for 2026—a 30% year-over-year increase—reinforcing its dominance in Taiwan and global ventures.

PC prices are set for significant hikes starting March 2026 due to severe memory shortages. Lenovo warns partners of impending increases, with PC DRAM projected to surge 105-110% in Q1 2026. AI data center demand is driving this industry-wide shift.

The PC sector is bracing for significant price increases starting early March 2026, driven by a persistent global memory chip shortage. Lenovo's North America channel chief, Wade McFarland, alerted partners in a February 2 letter that select commercial client devices will see price adjustments. He urged partners to place orders by February 25 (through distributors by February 28) to beat the upcoming update, noting that orders received but not shipped by March 31 will be repriced. This industry-wide shift stems from the ongoing AI-driven data center boom, which has sharply pushed up DRAM and NAND prices. Major OEMs, including Dell, Lenovo, HP, and HPE, are set to implement these hikes. Even Lenovo, the world's largest PC maker, with component stocks reportedly 50% above normal, isn't fully shielded. While CFO Winston Cheng indicated enough memory chips for the full year, the company aims to balance pricing with availability throughout 2026. Blended DDR4 and DDR5 prices are expected to jump an astounding 105–110% in Q1 2026, significantly outpacing server DRAM, which is forecast to rise 88–93%.

Applied Materials projects robust Q2 revenue of $7.65B, exceeding analyst consensus, despite a $252M penalty. This resilience stems from surging HBM demand for AI and strong DRAM growth via 3D chiplet stacking. China remains its top market, contributing 30% of Q1 revenue.

Applied Materials demonstrates strong financial resilience, forecasting fiscal Q2 revenue of approximately $7.65 billion. This figure significantly surpasses analysts’ consensus of $7.03 billion, even after accounting for a $252 million penalty related to illegal equipment shipments to China. CEO Gary Dickerson highlights High-Bandwidth Memory (HBM), critical for AI computing, as a primary growth catalyst. The company expects its semiconductor systems business to expand over 20% by 2026. DRAM is also projected as the fastest-growing segment in 2026, driven by increasing adoption of 3D chiplet stacking – a key packaging technology for AI processors. Memory equipment demand, including etch and deposition tools crucial for DRAM, is experiencing a boom as major customers like Samsung Electronics and Micron Technology ramp up capacity to address tight supply. Financially, Applied Materials anticipates Q2 earnings around $2.64 per share, ahead of the $2.29 consensus. Q1 revenue dipped 2% year-over-year to $7.01 billion (a smaller decline than feared), with earnings at $2.38 per share, exceeding projections of $2.21. Notably, China remains a cornerstone market, accounting for 30% of 1QFY26 revenue—its largest regional share—outperforming Taiwan (25%) and Korea (21%). This continued presence persists despite the recent penalty and ongoing U.S. export curbs, which previously led the company to caution about a potential $600 million reduction in fiscal 2026 revenue and plans to cut roughly 4% of its global workforce.

Medtronic's Hugo RAS debuted in the U.S. with a successful prostatectomy by Dr. Kaouk at Cleveland Clinic. Patient discharged next day. This marks Medtronic's challenge to existing soft-tissue surgical robotics, expanding minimally invasive access and offering surgeons a new choice.

Medtronic's Hugo RAS system has officially commenced its first U.S. commercial cases. Dr. Jihad Kaouk, MD FACS, FRCS (G), Cleveland Clinic’s robotic surgery chair, successfully performed a robotic-assisted prostatectomy, with the patient discharged just one day post-surgery. Dr. Kaouk highlighted this as a significant step towards expanding access to minimally invasive surgery and customizing patient approaches. This follows Hugo's FDA clearance for urologic procedures late last year, marking Medtronic's strategic entry into the U.S. soft-tissue surgical robotics market, long dominated by a single competitor. Rajit Kamal, Medtronic VP, emphasized Hugo's potential to enable more patients to access minimally invasive benefits. CEO Geoff Martha expressed strong confidence, outlining plans to expand into additional indications in the U.S. like hernia, part of broader general surgery, where Hugo "really shines." Medtronic is uniquely positioned as the only company globally offering approved solutions across open, laparoscopic, and robotic-assisted surgeries. This comprehensive offering is highly valued by surgical practices and hospitals such as Duke University Hospital and Atrium Health Wake Forest Baptist High Point Medical Center, which are among the first to install the Hugo RAS system. Hugo offers surgeons a crucial "another choice" in advanced surgical technology.

Two handset makers are testing 7-inch smartphones, like the 6.96-inch Huawei Mate 70 Air. This aims to counter homogenization, offering more internal space for batteries/cameras (+0.3 inches) and improved 6.7-to-7-inch viewing comfort. Challenges include portability, but new models like the vivo X300 Max are expected.

The market is buzzing with reports that two handset makers are currently testing 7-inch large-screen smartphones, hinting at a potential revival for these devices. This move comes as manufacturers seek to differentiate their products in an increasingly standardized market. Indeed, 7-inch smartphones aren't entirely new. Huawei launched the Mate 70 Air last year, a large-screen device measuring 6.96 inches with a striking body width of 81.5mm, compared to the Xiaomi 17 Ultra's 77.6mm. Renewed interest in larger displays is primarily driven by two key factors: 1. Enhanced Internal Space: Even an additional 0.3 inches of screen can increase internal volume, allowing manufacturers to integrate bigger batteries, more powerful camera systems, and improved cooling solutions. 2. Superior User Experience: Moving from a 6.7-inch display to a 7-inch one brings a noticeable improvement in reading comfort and overall video immersion. This strategic shift can be seen as an attempt to combat product homogenization. Mainstream flagship models are becoming increasingly similar in size, limiting differentiation. While foldable phones have entered the market, they haven't addressed all large-screen demands, as they require an extra step of unfolding. A 7-inch phone, however, offers a larger display by default. The previous decline of large-screen "bar" phones occurred when advancements in screen and bezel technologies made 6.7 inches the norm, shifting big displays from a selling point (when 5.5 inches was considered "giant") to a standard feature. Despite the appeal, portability remains a core challenge. An oversized phone risks feeling like a "small tablet," sacrificing the convenience of a smartphone. Manufacturers must address weight distribution, balance, center-of-gravity design, and ensure the operating system fully leverages the larger display. Looking ahead, Huawei is likely to release an updated version of its Mate 70 Air. vivo has confirmed plans for a model called the vivo X300 Max, with "Max" widely signaling a larger display. Additionally, the HONOR 500 series is expected to introduce an Ultra variant with a bigger screen size.

SK hynix (14.4 Gbps, 1c 10nm) and Samsung Electronics (12.8 Gbps, 12nm, 21% better efficiency) are set to unveil LPDDR6 at ISSCC 2026. Samsung's LPDDR6X samples are already with Qualcomm for its 2027 AI250 accelerator, signaling rapid next-gen adoption.

While HBM4 commands attention, a fierce LPDDR6 memory race is now unfolding. SK hynix and Samsung will showcase their next-generation LPDDR6 memory at ISSCC 2026 in San Francisco, from February 15–19. SK hynix is pushing 16Gb LPDDR6 modules at an impressive 14.4 Gbps. These modules leverage the company’s 1c 10nm DRAM process, marking its 6th generation of 10nm DRAM, nearing the current JEDEC standard limits for LPDDR6. Samsung’s LPDDR6 chips, upgraded since CES 2026, are designed for 12.8 Gbps performance for 16Gb modules, a notable boost from their initial 10.7 Gbps versions. Despite utilizing a slightly larger 12nm process compared to SK hynix's 10nm node, Samsung's LPDDR6 still reports a significant 21% improvement in energy efficiency over its predecessor, LPDDR5X. Intriguingly, LPDDR6X, the subsequent generation, is already attracting major interest. Samsung has reportedly provided samples of LPDDR6X to Qualcomm. This unusually early move is attributed to Qualcomm’s tight chip development schedule, aligning with their work on next-generation semiconductors for servers and automotive applications. Qualcomm plans to deploy LPDDR6X in its AI accelerator, AI250, slated for release in 2027. Commercial production for LPDDR6X is anticipated in the second half of 2027.

SECO and MediaTek will unveil their Edge AI solutions at Embedded World 2026 (March 10-12). Solutions like the SOM-Trizeps-X-Genio360/360P deliver >8 TOPS AI performance under 5W for cost-efficient systems. The Modular Vision Genio700 HMI showcases local AI for ultra-low latency medical automation.

At Embedded World 2026, SECO and MediaTek will showcase their advanced Edge AI solutions for diverse industries. Introducing SOM-Trizeps-X-Genio360/360P: SECO expands its Trizeps SODIMM family with new SOMs based on the recently launched MediaTek Genio 360 and 360P IoT platforms. These 6nm, value-tier processors balance strong CPU with power efficiency. Their integrated NPU delivers over 8 TOPS AI performance, enabling on-device and Generative AI at the edge. Compact (67.6x36.7mm), these SOMs support 4K displays, camera interfaces, and industrial connectivity. With <5W power and -40°C to +85°C industrial temp range, they target HMIs, mobile devices, smart appliances, and AI edge systems. SECO CTO Davide Catani highlights the Genio 360-based SOM's cost/power efficiency, accelerating on-device AI time-to-market. MediaTek's CK Wang notes their chips' "fantastic AI performance, premium connectivity, and unparalleled power efficiency" for Industrial IoT. Modular Vision Genio700 - AI-Powered Industrial HMIs: SECO's 15.6-inch Modular Vision Genio700, powered by MediaTek Genio 700, offers high-performance local AI for medical/intelligent apps, without cloud dependency. A Genio 510 variant is also available for optimized performance. Live demos at Embedded World (SECO & MediaTek booths) will showcase medical automation. See local AI models (face recognition, voice commands, sample inspection) run with ultra-low latency, integrated with SECO’s Clea for secure device management. Unified Software Stack for Edge AI: SECO’s MediaTek solutions feature a unified software stack, accelerating edge AI. It integrates SECO AI-Box applications with Developer Center resources.

Intel, Samsung, and SK hynix accelerate 1.4nm chip production with ASML's $380M High-NA EUV, enabling 8nm feature resolution by 2027-28. This contrasts with TSMC's decision against High-NA for 1.4nm, citing escalating costs. A high-stakes strategic divide emerges in the semiconductor race.

The race to 1.4nm chips intensifies as ASML prepares its next-generation High-NA EUV lithography machines for mass production by 2027-28. These tools boost numerical aperture (NA) from 0.33 to 0.55, resolving 8nm features (vs 13nm), critical for 1.4nm logic and 10nm-and-beyond DRAM. Intel deployed ASML's Twinscan EXE:5200B in Dec 2025 for its 14A process. Samsung followed, receiving its first High-NA EUV late 2025, a second in H1 2026, for 2nm lines (including Exynos 2600 and Tesla AI chips). Japan's Rapidus targets 2nm mass production H2 FY2027 (starting at 6,000 wafers/month, scaling to 25,000) and 1.4nm by 2029. In memory, SK hynix installed High-NA EUV in Sept 2025, planning five EUV layers for its 1c DRAM. Micron also boosts EUV-ready lines. However, TSMC, the world's largest foundry, opts not to adopt High-NA EUV for its 1.4nm node. Each unit costs ~$380 million, adding to already high wafer prices ($30,000 for 2nm). This strategic divergence highlights a core question: will cutting-edge tools or established process expertise win the silicon future?

China's AI race intensifies. DeepSeek expanded its context window tenfold to 1M tokens, while Zhipu AI's GLM-5 debuted with enhanced capabilities. Zhipu's 30% price hike signals a shift from price wars to technical strength, backed by fierce competition from Alibaba and ByteDance in text and video AI.

China's AI model race is rapidly accelerating. DeepSeek, a Chinese AI startup, significantly upgraded its flagship model, expanding its context window from 128,000 to over 1 million tokens – a nearly tenfold increase. This bolsters its ability to process complex prompts. Its knowledge cut-off was also extended from July 2024 to May 2025, adding almost a year of new data. However, it remains text/voice focused, lacking multimodal visual understanding. This DeepSeek upgrade, ahead of anticipated DeepSeek V4, coincided with Zhipu AI's GLM-5 debut. GLM-5 offers enhanced coding and agentic capabilities, driven by a doubled parameter count and DeepSeek Sparse Attention for efficiency. Market interest is strong: Zhipu AI increased its GLM Coding Plan price by 30% this week. This suggests a shift from prior price-based competition among Chinese models to clear improvements in technical strength and market competitiveness. GLM-5 is also optimized for deep inference across major Chinese computing platforms like Huawei's Ascend, Moore Threads, and Cambricon, ensuring broad deployment. The rivalry extends beyond text. Bloomberg notes Alibaba is set to unveil its Qwen-3.5 model. ByteDance (TikTok's parent) is testing Seedance 2.0, a video-generation model positioning itself as an OpenAI Sora rival, already producing striking demo videos.

Zimmer Biomet commercially launches its Phantom Curved TTC Nail, stemming from its $1.2B Paragon 28 acquisition. This anatomic design improves soft tissue preservation, torsional stability, and alignment reproducibility for complex foot fusions, outperforming traditional straight nails and reaffirming Zimmer's strategic growth.

Zimmer Biomet has announced the commercial launch of its Phantom Curved TTC Nail system, expanding its foot and ankle portfolio. This next-generation system is a direct outcome of Zimmer's $1.2 billion acquisition of Paragon 28 last year, reinforcing its commitment to maintaining Paragon 28’s established business structure and pre-acquisition growth, a strategy highlighted by Zimmer CFO Suketu Upadhyay. The Phantom system's curved design directly follows the natural anatomic arc of the foot. This represents a significant advancement over traditional straight tibiotalocalcaneal (TTC) arthrodesis nails, which often rely on non-anatomic trajectories. Albert DaCosta, Zimmer Biomet Global President, emphasized that the system helps surgeons "achieve reproducible alignment, strong fixation, and thoughtful soft-tissue preservation" in challenging fusion cases. Key advantages of the curved implant include enhanced soft tissue preservation, increased torsional stability, and superior alignment reproducibility. Its design extends precisely from the posterior calcaneus through the subtalar joint, into the talus, and along the tibial metaphysis. The system is suitable for a wide range of procedures, including primary and revision TTC fusion, complex Charcot reconstruction, post-traumatic cases, and rectifying failed straight nail fusions.

Samsung began mass production of HBM4 AI memory chips on Feb 13, 2026, claiming an "industry-leading" breakthrough. These chips are >40% faster than previous models, vital for AI data centers, and position Samsung as a front-runner in a memory market projected to exceed US$840 billion.

Samsung Electronics officially announced on Feb 13, 2026, the mass production and commercial shipment of its next-generation high-bandwidth HBM4 memory chips. This achievement marks a "first in the industry," securing an early leadership position critical for scaling the vast data centers powering the AI explosion. The new HBM4 chip demonstrates significantly faster processing speeds, exceeding industry standards by over 40 percent. This enhanced performance directly addresses the escalating demands for higher performance in AI applications. US tech giant NVIDIA is widely expected to be a primary customer. The HBM4 race is intense. While Samsung and South Korean rival SK hynix Inc. are leading producers, Samsung struggled with HBM3, lagging behind SK Hynix. However, with this early HBM4 production, Sejong University business professor Kim Dae-jong notes Samsung has positioned itself as a "front-runner." Competitor Micron Technology, faced a SemiAnalysis report suggesting it might be left off Nvidia's HBM4 supplier roster. However, Micron's CFO Mark Murphy rebutted this, stating they are in high-volume HBM4 production, shipping to customers, and their product delivers performance exceeding 11 gigabits per second, with volumes ramping up in Q1 – one quarter earlier than previously announced. The global memory industry is poised for significant growth, with Taipei-based TrendForce Corp. predicting revenue to surge past US$840 billion next year. Samsung has already committed billions to expand chip production facilities, focusing on advanced manufacturing and upgrading existing lines to meet this rising demand.

Firefly's Alpha Flight 7 completed a 20-second static fire, targeting a Feb. 18, 2026 launch. This test flight, the last in its current form, validates key Block II upgrades like enhanced thermal protection systems and stronger composite structures, paving the way for future mission reliability and performance.

Firefly Aerospace has successfully completed a full-duration 20-second static fire at its SLC-2 launch complex at Vandenberg Space Force Base. This milestone prepares Alpha Flight 7 for its anticipated launch no earlier than Feb. 18, 2026. Flight 7 is the final mission in Alpha's current configuration. It serves as a pivotal test to achieve nominal first and second stage performance and, crucially, to validate key systems ahead of Firefly's Block II configuration upgrade. The Block II upgrade is designed for enhanced reliability and manufacturability, featuring a 7-foot increase in Alpha's length, consolidated in-house batteries and avionics, an enhanced thermal protection system (TPS), and stronger carbon fiber composite structures built with automated machinery. Testing these specific subsystems, including the in-house avionics and thermal improvements, on Flight 7 will enable Firefly to gain vital flight heritage and validate lessons learned, ensuring a seamless transition and enhanced performance for the full Block II deployment.

Cirrus G3 Vision Jet's Composite Revolution & 7-Seat Leap Quicktake: Cirrus G3 Vision Jet debuts an advanced 7-occupant cabin and 30+ refinements, leveraging carbon fiber composites for enhanced durability and space. Backed by award-winning safety (CAPS, Safe Return) and a history of 700+ deliveries, it redefines personal aviation efficiency and peace of mind.

The Cirrus G3 Vision Jet advances personal aviation, building on extensive composites and safety expertise. Its entire airframe, including fuselage and wing, is a seamless carbon fiber composite monocoque. This design boosts structural integrity, durability, and expands passenger cabin space. The G3 features a reimagined interior with premium materials and expanded seating for up to seven occupants (six adults, one child) in Premium/Arrivée trims, enabling versatile missions. It incorporates over 30 refinements, including ATC Datalink. Core safety systems like the Cirrus Airframe Parachute System (CAPS) and Safe Return Emergency Autoland provide peace of mind. The Perspective Touch+ flight deck minimizes pilot workload with automatic database updates, 3D SafeTaxi, and new Spectra wingtips/lights. CEO Zean Nielsen emphasizes "relentless innovation" behind the G3. The Vision Jet's success is evident: prototype flew in 2008, FAA certified 2016, and became the most-delivered business jet within two years of its Dec 2016 launch. By Dec 2024, 600 units were delivered (avg. 75/year). Over 700 Vision Jets have been delivered by 2026, showcasing its market impact.

Chiplets introduce extreme thermal-mechanical stresses, pushing defectivity targets lower (e.g., from 10 DPPM to <1 DPPM per chiplet). New materials, assembly forces (millions of solder bumps), and longer interconnects demand holistic design and evolving EDA tools to prevent costly field failures.

The rapid adoption of chiplet architectures in data centers escalates reliability concerns. Assembling multiple chiplets into a package introduces severe thermal-mechanical stresses, leading to warpage, cracking, and long-term failures. A 10 DPPM system target necessitates <1 DPPM for each chiplet in a multi-chiplet package, fundamentally altering 2D design approaches. These stresses arise from diverse sources: manufacturing pressure (e.g., millions of solder bumps creating significant force), differing coefficients of thermal expansion (CTEs), and complex 3D-IC interconnects like TSVs. Issues once handled by foundries (e.g., materials choices) now impact design teams, breaking down traditional silos. EDA tools are evolving for advanced stress analysis and interface verification. Debugging 3D-ICs faces challenges from innovations like backside power delivery (e.g., Intel 20A), which limits traditional Focused Ion Beam (FIB) methods. Expensive 2nm nodes and 3D-ICs require more granular redundancy strategies beyond simple core duplication, unlike current server-class approaches. While chiplets simplify individual functions (e.g., enabling reliable 12nm analog circuits), system-level integration remains complex. With interconnect distances up to 25mm (e.g., UCIe) and no current standardization for chiplet boundaries, robust, package-level verification from architectural stages is critical. This holistic approach is essential to avoid costly field failures and enable a future chiplet marketplace.

Tower Semiconductor is rerouting 300mm wafer production to Japan after Intel withdrew from their 2023 New Mexico agreement. Amidst this, Tower reported 2025 revenue of $1.6B (up 9%) and forged a key silicon photonics partnership with NVIDIA for 1.6T AI optical modules.

Tower Semiconductor faces a production shift as Intel intends to withdraw from a September 2023 agreement to manufacture 300mm wafers in New Mexico. This deal, valued with Tower's planned $300M equipment investment for over 600,000 photo layers per month, followed Intel's failed $5.4B acquisition of Tower due to China's regulatory clearance. Production volumes are now rerouted to Intel's Fab7 in Japan, with mediation ongoing. Despite this strategic pivot, Tower shows momentum. The company posted Q4 2025 revenue of $440M (up 14% year-over-year) and full-year 2025 revenue reached $1.6B (up 9%), with net income at $220M (up 6%). Q1 2026 guidance projects $412M in revenue, a 15% YoY gain. Crucially, Tower is teaming with NVIDIA for next-gen AI infrastructure, developing 1.6T optical modules for data centers. This partnership leverages Tower's advanced silicon photonics (SiPho) platform. Their SiPho technology underpins 1.6T optical modules supporting NVIDIA's networking standards, offering up to twice the data rate of previous generations, boosting bandwidth density and throughput for AI-driven optical links.

Montage Technology, a key memory interconnect chip supplier, saw its Hong Kong IPO soar 57% on Feb 9. Backed by JP Morgan and Alibaba, it dominates over one-third of the global market, with 2024 revenue reaching RMB 3.64B, fueling AI data center expansion.

Shanghai-based Montage Technology, a designer of chips vital for data centers and AI accelerators, made a significant splash with its Hong Kong debut on February 9th. Shares surged 57% from its HK$106.89 price to HK$168, backed by cornerstone investors including JP Morgan and Alibaba. Montage holds a leading position, ranking as the world’s largest memory interconnect chip supplier in 2024, commanding over one-third of global revenue. Its PCIe retime and CXL MXC (Memory eXpander Controller) chips are crucial for accelerating data flows, serving key customers like Intel, AMD, and Samsung, directly supporting the global AI data center buildout. Financially strong, Montage reported RMB 3.64 billion in revenue and RMB 1.34 billion in net profit for 2024. Projections estimate 2025 net income at RMB 2.15–2.35 billion, potentially reaching RMB 3.3 billion in 2026. The IPO saw immense global demand: its public tranche was 707.3 times oversubscribed, and the international placement 37.67 times covered. This "A+H" dual-capital platform move, following its 2019 STAR Market listing, highlights a trend of Chinese AI and semiconductor firms entering capital markets, despite its HK opening price being a 44% discount to its Shanghai stock (RMB 170.90).

AI demand is powering foundries through a traditional slow season. TSMC projects record Q1 revenue ($34.6B-$35.8B), up 4% QoQ. Vanguard and PSMC are implementing significant price hikes (10-15%) due to full capacity and AI-driven tight supply, signaling robust industry growth.

Traditionally slow, Q1 2026 is seeing foundries thrive, driven by strong AI demand and recovering panel driver IC orders. TSMC leads, forecasting record Q1 revenue of US$34.6B-$35.8B, up 4% QoQ, fueled by advanced AI applications, making them Q1's top-performing foundry. Vanguard (VIS) reports strong Q1 momentum. Server power management ICs robust, with display driver IC demand recovering (TV/e-reader restocking). Wafer shipments expected 1-3% QoQ growth. Despite 3-5% ASP decline from product mix, VIS is at full capacity. After initial Q1 hikes, VIS notified customers of a second, 10-15% price increase from April, set to boost gross margins. PSMC also raising prices. Strong AI server-driven memory demand since January crowds out logic capacity. 12-inch DDIC and CIS wafer prices, from a low base, are rising since January. Increased power device (e.g., MOSFETs) demand for AI/edge computing prompts PSMC to raise 8-inch wafer foundry prices from March. UMC maintains stable Q1. Wafer shipments and ASPs flat, a stronger outcome than its typical 8-9% QoQ Q1 decline. Gains in smartphone ISP and AMOLED driver ICs support 22nm process growth.

Hexagon Purus divested its aerospace business, Hexagon Masterworks Inc., to SpaceX for $15M. This strategic move, involving composite storage cylinders for aerospace applications, bolsters Purus's finances and refines its portfolio, driven by re-evaluating North American hydrogen mobility potential.

Hexagon Purus (Oslo, Norway) announced on Feb. 5 the sale of its aerospace subsidiary, Hexagon Masterworks Inc., to Space Exploration Technologies Corp. (SpaceX, Hawthorne, Calif.). The deal encompasses 100% of Masterworks' shares, a producer of high-pressure composite storage cylinders for aerospace and space launch applications in North America. Critically, Masterworks' hydrogen mobility contracts are excluded and will transfer within Hexagon Purus. The transaction boasts an enterprise value of approximately $15 million, comprising $12.5 million cash at closing and a $2.5 million contingent earn-out. Hexagon Purus cites the aerospace business's need for a dedicated industrial owner to support its future growth. Furthermore, the company does not anticipate significant near-to-medium term potential in the North American hydrogen mobility market, making this divestment a strategic alignment with its ongoing portfolio review. This move strengthens Hexagon Purus’s financial position and extends its liquidity runway. Philpott Ball & Werner LLC, TCF Law Group, PLLC, and Advokatfirmaet Schjødt advised on the deal.

Samsung Display's Multi-Million-Dollar Bet Quicktake: Samsung Display is investing in A4 plant OLED capacity for Apple's foldable iPhones, anticipating increased demand. This move supports Apple's entry, projected to raise foldable phone penetration from 1.6% (2025) to over 3% by 2027.

Samsung Display is reportedly evaluating significant investment to expand its OLED production capabilities for Apple's upcoming foldable products. This includes potential upgrades at its A4 plant in Asan, South Chungcheong Province, particularly focusing on backplane-related processes like thin-film transistor (TFT) lines. Initial capital spending is anticipated to commence in Q2 after final internal review. This strategic expansion reflects Apple's growing expectations for the foldable category, with plans for a first-generation foldable iPhone and another model expected next year. Both devices are likely to be sold concurrently by 2027, significantly increasing foldable iPhone shipments and, consequently, demand for Samsung Display's OLED panels. Samsung Display expects to begin mass production of foldable OLED panels for Apple in June. Its A3 facility already has a dedicated Apple foldable OLED line with an annual capacity of approximately 15 million units (based on 7-inch class panels). Apple's first foldable phone, the "iPhone Fold," is expected to feature an in-folding design with a 7.58-inch internal and 5.35-inch external display. Around 10 million units are slated for production in its first year, with Samsung Display expected to remain the sole OLED panel supplier for several years. Beyond OLEDs, Apple is also evaluating transparent polyimide (PI) film for the cover window's protective layer, with South Korea's Kolon Industries, Inc emerging as a potential supplier. Apple's entry into the foldable market is a major catalyst: Industry analysts project foldable phone penetration will jump from 1.6% in 2025 to over 3% by 2027, driven by Apple's anticipated launch in H2 2026.

China's memory tech gap with South Korea is significantly wider than assumed, exceeding 5 years. Veteran DRAM expert Shim Dae-yong (26 years at SK Hynix) highlights restricted EUV access and dependency on key HBM materials (e.g., Japanese suppliers) as critical barriers to catching up.

Industry veteran Shim Dae-yong, a professor at Dong-A University and former SK Hynix core DRAM and early HBM development leader (26 years experience), asserts that China's technological lag in advanced DRAM is widely underestimated. He believes the gap with South Korean giants like Samsung and SK Hynix is not 2-3 years, but exceeds five. A primary challenge for China's advanced DRAM development is restricted access to EUV lithography. These tools are essential for producing DRAM at 10nm-class nodes and beyond. While multi-patterning can extend legacy lithography tools for some time (as Micron Technology did at the 1a node), EUV becomes indispensable from the 1b node onward. Chinese firms also grapple with yield hurdles, needing around 80-90% for commercially viable cost competitiveness. While China explores advanced 3D stacking as an alternative to bridge the gap without EUV, this approach merely shifts the bottleneck to materials and advanced packaging. Key HBM materials, such as underfill and epoxy molding compounds, are largely controlled by Japanese suppliers like Resonac and NAMICS Corporation. Even Samsung Electronics and SK hynix have spent years working to localize these critical inputs. Furthermore, Chinese memory makers lack the deep, collaborative experience South Korean firms have cultivated over decades with major technology partners such as Microsoft, Google, Apple, and NVIDIA. These partnerships are crucial for co-developing and validating memory products across new system architectures, a process not easily replicated in a short period.

Eric Knudsen is appointed ZOLL Medical Corporation CEO from April 1, succeeding Jon Rennert (10-year tenure). Rennert, who expanded Zoll's defibrillation and portfolio, becomes Executive Chair. Knudsen brings 30 years' medtech experience, including 6 years leading Zoll's Corporate Development, acquiring over a dozen technologies.

Effective April 1, Eric Knudsen will assume the role of CEO at Zoll, the renowned automated external defibrillator (AED) developer. He succeeds Jon Rennert, who transitions to Executive Chair after a transformative 10-year tenure as CEO. Rennert, who joined Zoll (an Asahi Kasei company) in 2016, expanded the company's reach in external defibrillation, advanced cardiopulmonary and respiratory solutions, and broadened its product portfolio. Knudsen brings nearly 30 years of extensive medtech industry experience, focusing on business development and corporate strategy. He previously led Zoll's Corporate Development team for six years, where he was instrumental in investing in strategic partnerships and acquiring over a dozen new technologies. Before his CEO appointment, Knudsen served four years as President of Zoll Itamar, a division specializing in diagnosing sleep disorders, having originally joined Zoll in 2013. His background also includes leadership positions at Covidien (now Medtronic) and Aspect Medical Systems. Zoll continues its commitment to innovation, advancing a wide range of products including wearable cardioverter defibrillators (WCDs) and monitor/defibrillator systems. Rennert expressed confidence in Knudsen's ability to drive "even greater growth and success," while Knudsen affirmed Zoll's dedication to "improving outcomes and helping save lives" globally with its medical devices and software.

Western Digital (WD) is seeing unprecedented demand, booking contracts to 2030. AI inference data explosion fuels robust Q2 FY26 growth: Net income up 210% to $1.8B, revenue up 25% to $3.017B. Projects >20% annual revenue growth, >50% gross margin, driven by critical HDD role.

Western Digital (WD), alongside Sandisk (spun off a year ago), is riding a significant wave in data and memory. CFO Kris Sennesael noted "all boats are rising" as AI workloads shift from training to inference, generating sharply rising data volumes. For cost-efficient, scaled inference, much of this new data will be stored on hard drives (HDDs), with hyperscale operators managing movement across SSDs, HDDs, and other tiers. WD has secured firm purchase commitments from its top seven customers through 2026. Furthermore, solid commercial agreements are in place with three of its top five customers—two extending through 2027 and another through 2028. Sennesael confirms WD’s 2026 production capacity is effectively fully booked. Discussions are actively underway to push contracts into 2028 or 2029, with some clients even inquiring about supply availability as far out as 2030. These long-term deals are structured around committed exabyte volumes and pricing. Financially, WD's Q2 FY2026 was strong: Net income soared 210% to $1.8 billion (from $594 million YoY), revenue climbed 25% year-over-year to $3.017 billion, and gross margin stood at 45.7%. The company projects over 20% annual revenue growth for the next three to five years, with gross margins above 50%. HDDs, storing roughly 80% of data in hyperscale cloud environments, are now a stable, AI-driven growth engine, moving past their historical cyclicality. WD is prioritizing higher drive density, with a roadmap aiming for a 100TB hard drive by 2029, powered by HAMR technology.

Apple plans a sub-$799 entry-level MacBook with an iPhone-class processor and 8GB DRAM. Despite rising memory costs, Apple mitigates risks via supply agreements, integrated design, and growing high-margin services (70-80% gross margin) and in-house silicon, targeting 5-8M units annually.

Apple is reportedly launching an entry-level MacBook priced below US$799, featuring an iPhone-class processor and 8GB DRAM – half the 16GB found in current models. This device, rumored to support Apple Intelligence, aims to expand market reach despite rising memory prices. Apple projects initial sales of 5–8 million units annually, equivalent to 20–30% of last year's Mac sales, with Quanta securing over half of the manufacturing orders. How is Apple confidently navigating surging memory costs? 1. Secured Supply: Apple has signed new DRAM and NAND supply agreements with major manufacturers like Samsung Electronics, SK hynix, and KIOXIA Group, ensuring supply through at least the first half of 2026. 2. Integrated Design Advantage: Apple's highly integrated hardware and software architecture allows for more efficient memory use, reducing overall DRAM requirements compared to Wintel laptops' open architecture, which needs additional memory headroom. 3. Services Growth: Apple's high-margin Services business (e.g., iCloud, App Store) boasts gross margins of 70–80%. Services revenue share grew from 19.8% in FY2022 to 26.2% in FY2025, providing a significant financial cushion against component price fluctuations. 4. In-House Silicon: Beyond memory, Apple designs key chips like M-series processors in-house. This strategy allows for economies of scale due to high volume, generating substantial cost savings that offset other component expenses. This strategic approach enables Apple to maintain profitability while introducing a highly competitive entry-level product.

Jupiter Endovascular concluded enrollment for its SPIRARE II pivotal trial (123 patients, 23 sites) evaluating the Vertex system. This trial assesses the novel TFX technology for acute, intermediate-risk pulmonary embolism (PE), aiming to reliably restore hemodynamics and support cardiac recovery.

Jupiter Endovascular has announced the completion of patient enrollment in its SPIRARE II pivotal clinical trial for the Vertex pulmonary embolectomy system. This prospective study successfully enrolled 123 patients across 23 sites in the U.S. and Europe, with Dr. Vikas Aggarwal at Henry Ford Hospital enrolling the final subject. SPIRARE II rigorously evaluates the Vertex system for acute, intermediate-risk pulmonary embolism (PE). Its endpoints characterize procedural and clinical performance, safety, right heart function, and clinical improvement up to 30 days post-procedure. The Vertex system incorporates Jupiter's transforming fixation (TFX) technology. TFX allows flexible, guidewire-delivered navigation, rapid saline pressurization for stable fixation during intervention, and a return to a relaxed state for repositioning or removal – designed for controlled and precise PE treatment. The Vertex catheter, integrating TFX, secured FDA investigational device exemption in August 2024, followed by FDA clearance in September 2025, and a subsequent $40 million funding round. Jupiter CEO Carl J. St. Bernard emphasized that this trial aims to evolve PE thrombectomy beyond mere thrombus debulking to reliably restoring hemodynamics and supporting cardiac recovery, demonstrating TFX technology's promise. Results are anticipated later this year.

Toray's new thermal welding process bonds thermoset/thermoplastic CFRP aircraft structures 3x faster than conventional methods. Tested on 900x600mm mock-ups, it delivers superior bonding strength (ISO 4587) and promises lighter, stronger airframes at aluminum-like production rates, transforming aerospace manufacturing.

Toray Industries Inc. has unveiled a groundbreaking thermal welding technology that bonds carbon fiber-reinforced plastic (CFRP) aircraft mock-up structures up to three times faster than traditional methods like adhesive bonding and bolted fastening. This innovation, first highlighted in February 2023, represents a significant leap for aerospace manufacturing efficiency. The successful tests involved a 900 x 600 mm aircraft mock-up. Thermoset CFRP, widely used for primary structures, formed the skin, stringers, and frames. Thermoplastic CFRP (CFRTP), ideal for high-rate production and complex geometries, was utilized for the clips. Toray's process efficiently welds these diverse materials. Leveraging extensive expertise in CFRP prepreg and molding, Toray’s thermal welding solution not only accelerates assembly but also delivers higher bonding strength compared to conventional adhesive bonding, as validated by ISO 4587 standards. This technology is a key outcome of the NEDO-supported "Development of New Innovative Composite Materials and Forming Technologies" project (2020-2024). The project aims to enable lighter, stronger CFRP aircraft with production rates matching or exceeding aluminum airframes. Toray is now actively accelerating commercialization with aerospace partners.

Siemens' Biograph One PET/MR Receives FDA Clearance. Siemens Healthineers' FDA-cleared Biograph One PET/MR merges simultaneous functional/metabolic imaging. It enhances personalized oncology by detailing tumor activity and response, and aids neurodegenerative diagnoses. Faster 50% scan times and AI integrations allow double patient throughput.

Siemens Healthineers has received FDA clearance for its Biograph One second-generation PET/MR scanner, integrating positron emission tomography (PET) and magnetic resonance (MR) for simultaneous visualization of organ location, function, and cellular metabolism. This advanced system significantly enhances theranostics, enabling more precise personalized oncology treatments by revealing tumor location, metabolic activity, and treatment response. It also emerges as a vital tool for neurodegenerative diseases like Alzheimer's, using PET to confirm diagnosis and MR to monitor related side effects. Building on the 2011 Biograph mMR, the Biograph One delivers substantial workflow improvements. It reduces patient scan times by 50%, allowing whole-body exams, which often exceed 60 minutes, to be completed in less than 30 minutes. This efficiency can support twice as many patients within typical radiology schedules. Its MR technology is based on the Magnetom Vida 3 Tesla scanner. The digital PET detector, featuring lutetium oxyorthosilicate (LSO) crystals (from the Biograph Vision PET/CT), boasts a 35 cm axial field of vision. This large detector reduces acquisition time for whole-body images by requiring fewer bed positions, increasing sensitivity for faster exams and flexible dose administration. The platform integrates several AI technologies. MyExam Companion simplifies workflows with advanced automation. Deep Resolve Boost 2D and 3D accelerates image reconstruction, removing noise and improving sharpness. Furthermore, flexible BioMatrix Contour XL MR coils provide effective whole-body patient coverage with fewer coils, reducing setup time and technologist radiation exposure. An integrated BioMatrix Position Sensor suggests optimal patient positioning, further streamlining exam setup. This convergence of advanced imaging and AI represents a shift towards more patient-centric pathways in personalized medicine, addressing the growing diagnostic challenges in cancer and neurological diseases.

Apple's 48-49% gross margin forecast faces rising memory costs (Samsung Electronics, SK hynix hikes) and TSMC's 3nm capacity limits (AI demand). Apple pivots by prioritizing premium 2026 iPhones (foldable, flagships) to maintain profitability, delaying standard models.

Apple is strategically navigating rising memory costs from suppliers like Samsung and SK hynix. While CEO Tim Cook noted a limited impact on Q1 gross margins, the effect is expected to be "more pronounced" in Q2. Despite forecasting strong 13-16% Q2 revenue growth and maintaining 48-49% gross margins, analysts remain concerned about longer-term profitability. Apple expects near-term margins to hold, driven by a richer iPhone 17 sales mix and growing services revenue. However, tight DRAM inventories through H2 2026 present significant uncertainty, fueling speculation on future iPhone price increases, a topic Cook declined to address directly. To counteract these pressures, Apple is reportedly prioritizing its three high-end iPhone models for 2026, including its first-ever foldable iPhone and two non-folding flagships with upgraded cameras and larger displays. This strategic shift aims to maximize revenue and margins from premium devices, delaying the standard iPhone 18 until H1 2027. However, memory isn't Apple's sole bottleneck. Cook highlighted that access to advanced-node manufacturing is the "primary factor" limiting iPhone output. Production is constrained by TSMC’s 3nm capacity for A-series and M-series SoCs. Demand for this advanced node has surged due to AI-related chips, intensifying pressure on available manufacturing slots. Apple's in-house chip strategy remains a competitive advantage, Cook confirmed, though its direct impact on offsetting rising memory costs wasn't quantified.

DeepSeek AI gained conditional approval for NVIDIA H200 chips, joining ByteDance, Alibaba Group, and Tencent, collectively authorized for over 400,000 H200s. Amidst U.S. scrutiny, NVIDIA awaits orders. DeepSeek also launches its V4 AI model mid-February, poised to rival GPT.

China's AI landscape is buzzing as DeepSeek, a leading startup, reportedly secured Beijing's conditional approval to acquire NVIDIA H200 AI chips. This aligns with its upcoming V4 AI model launch around the Lunar New Year. Reuters reports other tech giants like ByteDance, Alibaba, and Tencent are also authorized to collectively purchase over 400,000 H200s. While approvals are conditional, with specific terms being finalized by China’s National Development and Reform Commission (NDRC), NVIDIA CEO Jensen Huang confirmed no H200 orders from China yet, with shipments on hold. DeepSeek's potential H200 acquisitions face U.S. scrutiny, as a senior U.S. legislator alleged prior U.S. chipmaker assistance for AI models later used by the Chinese military. DeepSeek's V4 model, launching mid-February, promises enhanced coding capabilities rivaling closed-source models like Claude and GPT series. Jiemian notes V4's new mHC architecture and Engram conditional memory module aim to cut training costs and ease GPU memory bottlenecks, boosting China's local computing ecosystem despite domestic AI chips’ HBM limitations.

Airbus targets a 20-30% more fuel-efficient, 100% SAF-capable A320neo successor by the late 2030s. This "core priority" hinges on cost, sustainable, and industrial efficiency, including Open Fan engine tests (A380, 2029) to enable a ramp-up to 1,000 aircraft/year.

Airbus is outlining critical design and program priorities for its next-generation single-aisle (NGSA) aircraft, set to succeed the A320neo family in the second half of the 2030s. The NGSA aims for a significant 20-30% improvement in fuel efficiency over current models and will be engineered for up to 100% Sustainable Aviation Fuel (SAF) operation, marking it a "core priority" for Airbus's future commercial roadmap. As highlighted by Laurent Thomasson, Airbus coordinator for the European Clean Aviation research program, at the AIAA SciTech Forum, decision-making is centered on three key dimensions: Cost Efficiency: Ensuring the aircraft is affordable for airlines, passengers, Airbus, and its entire supply chain. Sustainable Performance: Focusing on reduced drag and optimized systems to minimize fuel burn. Industrial Efficiency: Crucial for supporting increased production. Airbus delivered 793 aircraft in 2025 (607 of which were A320neo family) and aims to ramp up to 1,000 aircraft per year. A central technical decision remains the propulsion system. Airbus is evaluating CFM International (CFM)’s Open Fan engine, alongside conventional high-bypass ducted options from Pratt & Whitney and Rolls-Royce, as the engine architecture heavily influences the airframe design. Plans include flight tests of the Open Fan concept on an A380 testbed under the Clean Aviation Take Off program. Readiness reviews are scheduled for 2028, with flight testing in 2029, targeting Technology Readiness Level (TRL) 6. Airbus emphasizes that certification planning must keep pace with technological maturation to ensure smooth industrialization of the NGSA. Beyond propulsion, Airbus is exploring advanced high-aspect-ratio wings with foldable tips, next-generation batteries for hybrid architectures, lightweight materials, and integrated systems designed to optimize operational and maintenance performance without requiring airport infrastructure changes.

ByteDance and Alibaba Group prepare to roll out new flagship AI models during the Lunar New Year holiday in February.

ByteDance is anticipated to launch three new models: Doubao 2.0 (a large language model), Seeddream 5.0 (image generation), and Seeddance 2.0 (video generation). Its Doubao chatbot has already achieved significant user adoption, reaching 163 million monthly active users (MAU) by December 2025. This makes Doubao China's largest AI application by user count, further boosted by its integration into the popular Douyin app. Alibaba plans its own counter-move with the release of its next-generation Qwen 3.5 model, also during the Lunar New Year. Qwen 3.5 is expected to demonstrate strong capabilities in mathematical reasoning and code generation, with specific optimizations for complex reasoning tasks. Alibaba will also initiate large-scale marketing campaigns for its consumer Qwen version to directly compete with ByteDance's Doubao. Beyond these immediate launches, DeepSeek's next-generation flagship model, DeepSeek V4—based on its new "MODEL1" architecture—is rumored for a mid-February 2026 release, indicating intensified future competition. While new players like Moonshot AI have emerged, releasing models like Kimi K2.5 (strong in code generation and visual data), they typically lack the proprietary cloud platforms and extensive user bases necessary to sustain large-scale model development. Tencent Holdings, though currently trailing ByteDance and Alibaba in AI model advancement, is actively pushing to catch up. In September last year, former OpenAI scientist Yao Shunyu joined Tencent as Chief AI Scientist, leading a significant reorganization of its AI teams.

Meta is doubling down on AI with a 2026 capex of up to $135B, driving demand for in-house ASICs like MTIA-2 (3nm, H1 2026) and MTIA-3 (H2 2026). This push is projected to boost ASIC-based AI servers to 27.8% market share by 2026, benefiting key supply chain partners.

Meta is aggressively investing in AI, projected to nearly double its 2025 capital spending to an estimated US$115–135 billion in 2026. This monumental investment directly fuels a surge in demand for AI servers and custom ASICs (application-specific integrated circuits), marking a significant strategic shift. At the heart of this strategy are Meta's in-house AI chips. The MTIA-2 is already in production and set for a H1 2026 debut. It leverages TSMC’s advanced 3nm process and CoWoS-S packaging technology, with Broadcom providing compute and I/O design services. Following swiftly, MTIA-3 is slated for a H2 2026 release, having taped out in Q3 2025. While also using TSMC’s 3nm process, MTIA-3 boasts a more complex design, featuring extra I/O and an additional SoC. Its larger reticle limits CoWoS packaging to just eight chips per wafer, leading Meta to partner with TSMC affiliate GUC for specialized back-end packaging and wafer processing. This aggressive ASIC development by Meta and other North American CSPs like Google is reshaping the AI server landscape. Industry analysts predict ASIC-based AI servers will constitute 27.8% of the market by 2026—the highest share since 2023. Overall 2026 AI server market growth will largely stem from North American CSPs, government sovereign clouds, and the expanding adoption of in-house ASICs and edge AI solutions. A notable ripple effect is the expected surge in Baseboard Management Controller (BMC) usage. Each MTIA chip is anticipated to utilize around 23 BMC units, creating a significant boon for suppliers like Taiwan’s ASPEED Technology.

Major chipmakers (Micron Technology, SK hynix, Samsung Electronics) are scrutinizing orders to combat hoarding as DRAM costs soar, impacting consumer electronics and PC shipments (9.4% decline forecast). Inventories are critically low (Samsung's DRAM at 6 weeks).

Memory tightness persists as leading chipmakers—Micron, SK hynix, and Samsung Electronics—intensify scrutiny of customer orders. They now demand end-customer details and volumes to prevent inventory hoarding and overbooking. This tightens supply, particularly for entry-level consumer electronics like TVs, smartphones, and PCs, with the automotive sector also facing significant pressure due to long qualification cycles. The financial ripple effects are clear: DRAM's share of a TV's Bill of Materials (BOM) cost has jumped from 2.5-3% to 6-7%, severely squeezing brand profitability. Consequently, TrendForce has revised its 2026 notebook shipment forecast, now predicting a steeper 9.4% year-on-year decline, up from the previous 5.4%. Manufacturers are adapting. Some salvage older memory chips for reuse in secondary markets, while PC makers design entry-level models with extra memory slots for future upgrades. Memory giants confirm the crunch: SK hynix reported sharp DRAM inventory drops and declining NAND flash stocks in Q4 last year, expecting continued tightness through H2. Samsung’s DRAM inventory has plummeted to just six weeks of supply, half the typical 10-12 weeks, signaling a supplier-driven market advantage for major producers.

Concurrent launches Kratos 32-core, the highest-core-count Intel Corporation Xeon 3U VPX board. It boosts performance by up to 60% over the 20-core version, setting a new standard for data-intensive, mission-critical edge computing.

Memory tightness persists as leading chipmakers—Micron, SK hynix, and Samsung Electronics—intensify scrutiny of customer orders. They now demand end-customer details and volumes to prevent inventory hoarding and overbooking. This tightens supply, particularly for entry-level consumer electronics like TVs, smartphones, and PCs, with the automotive sector also facing significant pressure due to long qualification cycles. The financial ripple effects are clear: DRAM's share of a TV's Bill of Materials (BOM) cost has jumped from 2.5-3% to 6-7%, severely squeezing brand profitability. Consequently, TrendForce has revised its 2026 notebook shipment forecast, now predicting a steeper 9.4% year-on-year decline, up from the previous 5.4%. Manufacturers are adapting. Some salvage older memory chips for reuse in secondary markets, while PC makers design entry-level models with extra memory slots for future upgrades. Memory giants confirm the crunch: SK hynix reported sharp DRAM inventory drops and declining NAND flash stocks in Q4 last year, expecting continued tightness through H2. Samsung’s DRAM inventory has plummeted to just six weeks of supply, half the typical 10-12 weeks, signaling a supplier-driven market advantage for major producers.

Glucotrack (NASDAQ:GCTK) secured three new U.S. patents for its innovative 3-year continuous blood glucose monitor. This device offers real-time, direct blood glucose readings with no on-body components, enhancing convenience and accuracy, positioning it closer to commercialization.

Glucotrack (Nasdaq:GCTK) has significantly strengthened its continuous blood glucose monitoring (CBGM) platform by receiving new U.S. patents from the U.S. Patent and Trademark Office (USPTO). The issued patents, US 12,453,494, US 12,458,257, and US 12,458,258, bolster the company's competitive position and intellectual property as it progresses toward commercializing its unique CBGM system. What sets Glucotrack's technology apart? Unlike traditional continuous glucose monitors (CGMs) that typically require an external on-body component and measure glucose in interstitial fluid (which can result in a lag time), Glucotrack's device features no on-body external component. It is engineered for an impressive three years of continuous, accurate blood glucose monitoring. By measuring glucose levels directly from the blood, it aims to provide real-time readings, promising a more convenient, less intrusive, and highly accurate solution for diabetes management.