01
SpaceX and Tesla plan USD 16.8 billion initial investment for Texas Terafab chip complex
AI Chip ManufacturingRegional CapacityTesting
SpaceX and Tesla plan an initial USD 16.8 billion investment to build the Terafab semiconductor complex in Grimes County, Texas. The project is designed to support AI chip demand for Tesla and SpaceX applications, with a long-term vision that could cover manufacturing, testing, and other chip-production stages.
The project reflects a growing trend: large technology companies are seeking deeper control over strategic AI compute supply, not only through chip design but also through manufacturing partnerships and regional production capacity.
R&A View: AI infrastructure is pushing major technology companies closer to the semiconductor supply chain. If more large end users pursue dedicated chip capacity or regional manufacturing models, supply planning may become more fragmented. Advanced packaging, testing, power delivery, high-density substrates, thermal materials, and manufacturing services could become more closely tied to specific customer ecosystems and long-term capacity commitments.
02
AMD acquires Taalas to strengthen AI inference accelerator roadmap
AI InferenceAcceleratorsSystem Architecture
AMD announced the acquisition of Taalas, a Toronto-based AI chip startup focused on silicon technology for AI inference workloads. Taalas develops technology aimed at reducing compute and memory bottlenecks during real-time model execution.
This shows that AI competition is moving beyond training performance. Inference efficiency, memory bandwidth, system cost, and workload-specific architecture are becoming more important as AI applications scale into real-time deployment.
R&A View: AI inference growth could drive demand for a wider range of semiconductor components beyond high-end GPUs. Memory bandwidth, power efficiency, interconnect design, accelerator modules, server boards, retimers, high-speed connectors, and thermal solutions may become more important in future AI infrastructure planning. Alternative accelerator ecosystems may also create new sourcing options, but qualification and platform maturity will remain key considerations.
03
Samsung launches next-generation AI memory technology with V10 Bonding V-NAND
AI MemoryV-NANDEnterprise SSD
Samsung introduced its latest AI-focused memory technology at the 2026 Future of Memory and Storage conference. The new V10 Bonding V-NAND uses more than 400 layers and a wafer-bonding architecture, improving storage density compared with the previous generation.
The update reflects the growing role of high-capacity NAND and advanced memory structures in AI workloads. As AI applications move from training to real-time interaction, storage density, bandwidth, energy efficiency, and thermal performance are becoming more important.
R&A View: Memory pressure is broadening from HBM into NAND, enterprise SSDs, and future memory-stack architectures. AI servers and storage-heavy applications may require closer tracking of NAND supply, controller availability, enterprise SSD lead times, thermal design, and long-term memory agreements. New memory technologies can improve future density and efficiency, but near-term availability still depends on ramp-up speed, qualification, and customer allocation.
04
Onsemi raises outlook as AI data-center power-chip demand accelerates
Power ICMOSFETAI Data Centers
Onsemi forecast third-quarter revenue above market expectations, supported by strong demand for power-management chips used in AI data centers. The company said AI data centers remain its fastest-growing business and expects related revenue to more than double in 2026.
This confirms that AI infrastructure demand is spreading into the power tree, not just processors and memory.
R&A View: AI data centers are increasing demand for power-management ICs, MOSFETs, drivers, protection devices, power modules, current sensing, and high-efficiency conversion components. As rack power density rises, component selection may become more constrained by thermal performance, efficiency, package type, reliability, and qualified supplier base. Power semiconductor availability should be reviewed alongside AI server deployment plans.
05
Microchip Technology forecasts stronger demand across AI data centers and industrial markets
MCUAnalog ICIndustrial
Microchip Technology forecast stronger-than-expected quarterly revenue and profit, citing demand for chips used in AI data centers as well as industrial, automotive, and aerospace applications.
This is important because Microchip’s product portfolio reaches many embedded and control-related applications, not only high-performance compute.
R&A View: Demand recovery in embedded and industrial semiconductors may become more selective. MCUs, analog ICs, interface chips, timing devices, power-management components, security ICs, and connectivity products could see stronger pull from AI infrastructure, industrial automation, automotive electronics, and aerospace systems. Long-lifecycle projects may need earlier checks on approved alternatives, date-code requirements, and continuity of supply.
06
GlobalFoundries tops quarterly expectations on data-center specialty chip demand
Specialty FoundryMixed SignalConnectivity
GlobalFoundries reported second-quarter revenue above expectations, supported by growing demand for chips used in data centers. The company manufactures specialty semiconductors for communications, data centers, automotive, industrial, and consumer electronics applications.
This highlights the importance of specialty process nodes in the AI infrastructure cycle. Not every AI-related component depends on leading-edge logic; many supporting devices rely on RF, mixed-signal, power, connectivity, and specialty technologies.
R&A View: Specialty foundry demand may become an important indicator for broader component availability. Data-center expansion can support RF, silicon photonics, mixed-signal, power-management, timing, interface, and connectivity chips. Even when leading-edge GPU supply receives most attention, specialty-node capacity and supplier allocation can still affect delivery schedules across networking, industrial, automotive, and data-center systems.
07
Infineon sees AI data-center power supply as a major growth driver
InfineonPower SupplyEnergy Efficiency
Infineon said power supply solutions for AI data centers remain in very high demand and continue to be its most important growth driver. The company now expects AI-related revenue to exceed its earlier fiscal-year forecast.
The signal is consistent with broader AI infrastructure pressure on power conversion and energy-efficiency components.
R&A View: AI data-center power demand is creating stronger opportunities for power semiconductors, controllers, MOSFETs, gate drivers, current sensors, protection devices, and high-efficiency conversion architectures. As more AI racks move toward higher power density, power-chain qualification, thermal headroom, and supplier capacity may become critical planning factors. Automotive and industrial recovery could further compete for some overlapping power and sensor components.
08
Sandisk and Western Digital highlight strong storage demand, while expectations remain high
StorageNANDEnterprise SSD
Sandisk and Western Digital reported strong results, supported by data-center demand and the broader AI-driven storage cycle. Expectations for storage and semiconductor companies remain high, even when earnings are strong.
The key supply-chain message is that AI-related storage demand remains strong, but the market is also sensitive to pricing trends, demand visibility, and how quickly capacity can catch up.
R&A View: AI workloads are increasing the importance of enterprise SSDs, NAND, controllers, high-capacity storage devices, and data-center storage infrastructure. Strong demand may support pricing and allocation pressure in selected storage categories. Projects requiring large storage capacity should monitor NAND supply, enterprise SSD qualification, controller availability, endurance requirements, and long-term sourcing options.
