Google is reportedly developing a new chip that embeds information from its Gemini AI models directly onto the silicon to boost efficiency. If the anticipated performance gains materialize, this new design would mark a deeper integration of the tech giant's...
Haesung DS has reportedly added China's CXMT as a customer for DDR5 package substrates as the South Korean supplier expands its DDR5 business and considers panel-based production for higher-layer-count products expected with the DDR6 generation.
Innolume GmbH has purchased Veeco Instruments' GEN2000 molecular beam epitaxy (MBE) system to expand quantum dot laser production for next-generation optical transceivers and AI-driven data center interconnects, addressing surging demand for co-packaged optics in hyperscale AI infrastructure. The investment, announced 21 July 2026, supports Innolume's ramp in manufacturing capacity for quantum dot (QD) diode lasers, which offer superior temperature stability, low power consumption, and high reliability—key requirements for future silicon photonics and co-packaged optics (CPO) architectures.
The GEN2000 MBE system, recognized for high throughput and low cost-of-ownership, will enable Innolume to meet growing customer demand as hyperscale AI and cloud computing drive optical connectivity needs. Market researcher LightCounting notes that CPO deployments are accelerating, with sales of optical transceivers using QD lasers projected to reach nearly $80 billion by 2031. QD lasers provide narrow spectral linewidth and low threshold current, making them ideal emitters for CPO.
Innolume CEO Sven Rüger stated that the GEN2000 provides the production capacity, material quality, and manufacturing efficiency required for their technology roadmap. Veeco’s VP & general manager for MBE, Matthew Marek, emphasized that the investment underscores the rising importance of MBE-grown QD lasers. Additionally, Innolume is installing Veeco’s Spector ion beam deposition system to expand in-house coating capabilities for ultra-low-loss anti-reflective and highly reflective laser facet coatings, strengthening end-to-end photonic device manufacturing. This dual adoption of Veeco’s MBE and ion beam platforms highlights complementary process technologies for high-volume production.
ABB acquires silicon carbide power conversion firm Advantics to expand DC portfolio for data centers, microgrids, and EV infrastructure.
Electrification and automation technology provider ABB (Zurich, Switzerland) is acquiring Advantics (Saint-Genis-Pouilly, France), a specialist in silicon carbide (SiC)-based power conversion solutions that integrate hardware, firmware, and software for demanding applications including data centers, industrial microgrids, power generation, and EV infrastructure. The acquisition, expected to close by Q4 2026, expands ABB’s direct current (DC) portfolio to meet accelerating demand for efficient DC solutions. The IEA forecasts electricity’s share of final energy consumption will rise from 20% today to about 36% by 2035. ABB states its DC solutions minimize repeated AC/DC conversion and maximize efficiency using advanced semiconductor technology. Advantics’ power converter modules deliver up to 99% efficiency. “ABB offers industry-leading DC solutions, and Advantics’ power converter technology and world-class engineering talent are a perfect match,” said Massimiliano Cifalitti, president of ABB’s Smart Power division. “Together, we will accelerate next-generation DC solutions for data centers, industrial microgrids, and EV infrastructure.” Advantics founder & CEO Michal Elias added, “As the world electrifies, demand for efficient, intelligent power conversion is accelerating faster than ever. ABB shares our vision.” ABB has over 140 years of history and around 110,000 employees. Its shares are listed on SIX Swiss Exchange (ABBN) and Nasdaq Stockholm (ABB).
TSMC expects "strong, multi-year" demand for AI chips as it ramps up its Arizona investment, signaling sustained growth in the semiconductor foundry sector. The Taiwanese chipmaker projects that surging adoption of artificial intelligence in data centers and edge devices will drive long-term orders, bolstering its expansion plans for the U.S. fabrication facility. The Arizona plant, part of TSMC's global capacity buildup, is positioned to serve major AI chip customers amid rising geopolitical pressures to diversify supply chains. This outlook aligns with broader industry trends, where AI chip demand is fueling investments in advanced nodes and packaging. The company's confidence in multi-year demand underscores the critical role of foundry capacity in powering AI workloads, reinforcing expectations for robust capital expenditure and revenue growth in the semiconductor ecosystem.
Keysight Technologies returns to the 2026 Design Automation Conference (DAC) as a Silver Sponsor and “I LOVE DAC” sponsor, showcasing its Design Engineering Software portfolio at Booth #927 from July 26-29 in Long Beach, Calif., targeting chip, package, RF/microwave, and quantum design engineers seeking EDA, AI-driven workflows, and advanced packaging solutions. The Keysight team will be on-site Monday through Wednesday (July 27-29) for live demos and one-on-one meetings, which attendees can book in advance via Keysight’s DAC event page. The company’s program addresses thorny chip and system design challenges, including accelerating agentic design workflows with surrogate modeling, managing and curating data for EDA AI workflows, and die-to-die signal integrity for advanced packaging. Advanced packaging engineers can explore chiplet, 3DIC, and 3D-HI interconnect challenges, including a poster on mixed-domain modeling for hatched ground planes validated on silicon bridge and mobile PCB test vehicles. Quantum computing gets its own spotlight in the “From Bit to Qubit: EDA Meets Quantum” panel, examining how quantum EDA platforms and digital twins intersect with chip design workflows. At Booth #927, Keysight will demo engineering data management tools, including SOS Enterprise for governed, AI-ready engineering data at scale, and RF Circuit Simulation Professional featuring an executable RF design workflow tool. Sessions and demos target chip, package, and RF/microwave design engineers, quantum hardware teams, and EDA leadership evaluating AI and governed data integration. Registration for DAC 2026 is free through the “I LOVE DAC” program.
General Compute, a fast inference neocloud building an ASIC-based alternative to GPU clouds, has secured a committed debt facility of up to $400 million from Upper90 Capital Management to scale one of the world's largest inference neoclouds, delivering AI results 16x faster than current GPUs. The financing starts at $100 million and scales with customer demand, marking one of the largest debt deals for a non-GPU cloud provider. General Compute holds no GPU allocation, freeing it to deploy specialized inference silicon from SambaNova (SN40 and SN50) while GPU clouds remain tied to a single chip supplier. The company has over $300 million in secured, price-protected supply and expects to be the first neocloud to deploy ASIC silicon at scale. Upper90, also an equity investor, aligns debt and equity behind the differentiated approach.
The announcement addresses the growing GPU dilemma: as Goldman Sachs predicts token consumption will rise 24x in 3.5 years, GPU clouds prove inefficient for inference, with new GPU racks demanding 120kW of power and specialized liquid cooling, trapping capacity behind multi-year data center builds. General Compute’s solution uses air-cooled ASIC chips requiring only 20kW per rack, no water cooling, and can be installed in weeks instead of years. Key performance claims include 16x faster inference, 7x faster time-to-first-token, 8.5x higher output throughput (1,000 tokens/second), and 6x better power efficiency than GPUs. The platform supports frontier models from OpenAI, DeepSeek, MiniMax, and others, with a switchover time of under 30 seconds. “We are the only neocloud that can serve premium tokens: frontier-level intelligence on the largest models, served fast,” said CEO Finn Puklowski, emphasizing the company’s lack of handcuffs to a single chip supplier.
CNT Firm Strengthens Executive Team for CNT Ecosystem
Texas Instruments Adds Cell-Level Intelligence to Battery Monitoring
Sorry, unusual traffic from your computer network.Your IP (107.174.184.84) has been blocked temporarily.To provide quality services to our members, DIGITIMES blocks IPs if large volume of traffic coming from a single IP address.If you see this page during normal browsing, please contact us and we will investigate your case. Please do not republish, publicly broadcast or publicly transmit content from this website without written permission from DIGITIMES Inc. Please contact us if you have any questions.
As the US continues to tighten export controls on advanced semiconductor equipment to China, questions are mounting over whether China can narrow the gap with leading-edge manufacturing through mature nodes and advanced packaging. German manufacturer Zeiss...
UMA: The Architecture Edge AI Needs to Scale
VSMC warns AI-ready MES is becoming essential for wafer fabs, with China accelerating adoption
19 h ago
TSMC's sustained capital spending highlights the two foundations of its competitive strength: technology leadership and manufacturing excellence. The latter remains a core advantage not only for TSMC but also for the broader semiconductor ecosystem built...
20 July 2026 Latest issue of Semiconductor Today now available For coverage of all the key business and technology developments in compound semiconductors and advanced silicon materials and devices over the last month, subscribe to Semiconductor Today magazine. Published today and free of charge, Semiconductor Today’s latest issue covers all that is new in gallium arsenide (GaAs), indium phosphide (InP), nitrides, silicon carbide (SiC), silicon germanium (SiGe) and other compound semiconductor materials. The magazine also covers the devices and applications that these materials enable. Click here or issue cover for direct access  Articles in this issue include: Monolithic all-GaN 2T1C-mLEDâphotodetector integration Purifying red InGaN micro-LED spectra Increasing passivated AlGaN power E-mode gallium nitride quantum well in aluminium nitride XHEMTs Silicon carbide technology patent activity remained strong in Q1 From road to rack: 800V EV innovations redefining AI data-center power architecture IVWorks, Aehr Test Systems, Aixtron, Ascent Solar, ASML, Atomera, AXT, Ayar Labs, BluGlass, ClassOne, Coherent Corp., Cree, Element Six, Elethron Ltd, Enkris, Guerrilla RF, Infineon Technologies, Innoscience, Indium Corp, IQE plc, Keysight Technologies, LayTec, Lumentum, Microchip Technology, Mitsubishi Electric, Navitas, Nimy Resources, Nexperia, NUBURU, onsemi, Oxford Instruments, Photon Design, Qorvo, Quantum Power Transformation Ltd, ROHM, SemiQ, Sivers Semiconductors, STMicroelectronics, Tower Semiconductor, US Critical Materials, Veeco Instruments, Volta Metals, WIN Semiconductors, Wise Integration, Wolfspeed, and much more. If your company is a supplier of equipment, materials, or services for the manufacture of compound semiconductor wafers and devices, and you are involved in marketing those products, then please see Semiconductor Today's 2025 Media Kit. Semiconductor Today is freely available to all. To learn more and to subscribe, visit: Subscribe for free.
Keysight Design Engineering Software at DAC 2026: Going Deep on Chiplets, RF and AI-Driven Design
19 h ago
Defacto Technologies, a 23-year-old chip design software company, will showcase an AI assistant and enhanced IP-XACT management at DAC 2025 (booth #1255, July 27–29, Long Beach), targeting SoC integration challenges for HPC, AI, automotive, and mobile markets. The core news is that Defacto’s SoC Compiler now leverages AI to automate pre-synthesis subsystem generation up to 100x faster, enabling even junior engineers to perform expert-level tasks. The company also addresses the industry’s shift to IP-XACT by offering seamless management alongside RTL, UPF, and SDC formats, reducing design complexity. Proven at major semiconductor companies for hundreds of projects, Defacto’s front-end integration platform delivers ROI by moving physically-aware design earlier in the flow, before synthesis. Headquartered in the French Alps with a US branch in California, Defacto provides 24/7 global support. At DAC, attendees can see the AI assistant, unified format handling, and new subsystem generation capabilities—solutions for design teams, RTL designers, SoC architects, and verification engineers tackling aggressive PPA requirements.
Nisshinbo Micro Devices Expands High-Voltage IC Lineup for Next-Gen Automotive 48 V Systems
Photonics Components for Future Unmanned Systems and Soldiers
Elon Musk has publicly rejected reports that SpaceX placed a roughly US$52 billion order with Foxconn for Nvidia GB300 AI servers, calling the widely circulated claim "fake news" in a post on X on July 20.
2026 WAIC. Credit: Leikeji China's smartphone industry is turning agentic AI into its next growth narrative, with ZTE's Nubia, StepFun and Honor presenting sharply different visions of how phones should plan tasks, coordinate apps and process personal data.
**Jilin University achieves record 2DEG mobility for N-polar GaN/AlGaN HEMTs on SiC via MOCVD, boosting high-frequency millimeter-wave performance.**
Researchers at Jilin University in China have reported a record two-dimensional electron gas (2DEG) mobility of 1947 cm²/V·s for nitrogen-polar (N-polar) GaN/AlGaN heterostructures grown on silicon carbide (SiC) substrates by metal-organic chemical vapor deposition (MOCVD) [Ma et al., *Appl. Phys. Lett.* 128, 072101 (2026)]. This is the highest value reported to date for N-polar GaN/AlGaN on SiC, addressing a key challenge in realizing high-performance N-polar GaN high-electron-mobility transistors (HEMTs) for W-band (75–110 GHz) applications, where output power densities up to 8 W/mm have been demonstrated.
N-polar GaN HEMTs offer advantages over metal-polar devices but suffer from rough surface morphology and high oxygen impurity concentrations (~10¹⁸ cm⁻³) that degrade mobility via interface roughness and ionized impurity scattering. Led by Yuantao Zhang and Gaoqiang Deng, the team varied the growth temperature of the high-resistance GaN (HR-GaN) template layer between 950 °C and 1000 °C. Counterintuitively, lowering the temperature increased surface roughness (RMS from 2.94 nm to 4.32 nm) but improved 2DEG mobility from 1468 cm²/V·s to 1947 cm²/V·s.
Secondary-ion mass spectrometry (SIMS) revealed that reducing the HR-GaN growth temperature increased carbon incorporation by an order of magnitude (from 1.1×10¹⁸ cm⁻³ to 1.1×10¹⁹ cm⁻³) while oxygen concentration dropped from 3.3×10¹⁷ cm⁻³ to 4.5×10¹⁶ cm⁻³. Under Ga-rich conditions, carbon preferentially occupies nitrogen sites, competitively suppressing oxygen incorporation. The reduced oxygen in the AlGaN barrier (from 2.1×10¹⁸ cm⁻³ to 2×10¹⁷ cm⁻³) weakens ionized impurity scattering—the dominant mobility-limiting mechanism. A semi-quantitative analysis using Matthiessen’s rule showed that a 10.5-fold suppression of ionized impurity scattering outweighs a 2.2-fold degradation from increased roughness.
Device-level impact was verified: N-polar HEMTs with HR-GaN grown at 950 °C achieved full pinch-off at VGS = −2 V and a saturation current density of 492 mA/mm at VGS = 1 V, while devices grown at 1000 °C could not be fully pinched off due to buffer leakage. The higher carbon concentration also increased HR-GaN sheet resistance from 7.8×10⁵ Ω/sq to 9.1×10⁸ Ω/sq, suppressing leakage current. The team concludes this work provides an effective approach for enhancing 2DEG mobility in N-polar GaN/AlGaN heterostructures, advancing high-performance N-polar GaN HEMTs.