Microsoft's in-house Azure Maia chips will become the main platform for its own AI services, including Microsoft 365 Copilot, GitHub Copilot and Microsoft Foundry. However, the company will keep using GPUs from Nvidia and AMD, Andrew Wall, general manager of the Azure Maia program, told DIGITIMES.

Microsoft will scale its in-house Azure Maia AI accelerators to multi-gigawatt deployments starting in 2027, making Maia the primary AI inference system for its products and services. The company also unveiled Maia 300, the third generation of the chip family, which it says will start production deployments next year.

Touch IC leader Elan Microelectronics recently announced its September 2026 consolidated revenue of NT$602 million (approx. US$18.9 million), down 18.74% month-over-month and 46.9% year-over-year. Third-quarter revenue slid 30.7% quarter-over-quarter to NT$2.221 billion (US$69.6 million), down 33.0% year-over-year, marking a sharp decline during what is traditionally a peak season.

The Financial Times recently reported that Nvidia is in early-stage talks to acquire or significantly deepen its investment in the US$25 billion open-weight AI startup Reflection AI. To dodge lengthy antitrust regulatory reviews, Nvidia is considering an "acqui-hire" framework to bring on key staff and license technology while expanding its strategic position in open-weight models.

As the global smartphone system-on-chip (SoC) market contracts, competition in the high-end application processor (AP) segment is intensifying. MediaTek is accelerating its push into the flagship market, putting greater pressure on Samsung Electronics, which is promoting 2-nanometer Exynos chips.

This week's (October 5 to October 11, 2026) coverage highlights intensifying efforts to build resilient semiconductor and AI hardware supply chains. Elon Musk's Terafab project faces manufacturing, packaging, printed circuit board (PCB) yield, partnership, and governance questions, while Intel remains involved and TSMC's role may be limited. Samsung reportedly advances HBM4E high-bandwidth memory validation and glass interposer testing. Vicor expands patent litigation over AI server power systems. Taiwan restricts China-made fiberglass sleeves, and China's new five-year plan prioritizes advanced PCBs, materials, equipment, and domestic investment.

Novatek reported stronger September and third-quarter 2026 revenue, suggesting seasonal smartphone demand and growth in edge AI chips are supporting the global display and chip supply chain. The results matter for international readers because they reflect broader shifts in consumer electronics demand, inventory patterns, and AI-related hardware spending.

Huawei consumer business group chairman Richard Yu said surging memory component costs have added more than US$200 to the hardware cost of each Huawei smartphone, putting fresh pressure on handset pricing as the broader memory market climbs rapidly.

MediaTek reported its September and third-quarter 2026 revenue results, with September revenue of NT$55.40 billion (approx. US$1.74 billion), down 13.69% from the previous month and up 1.97% from a year earlier. Third-quarter revenue reached NT$168.06 billion, up 10.43% quarter over quarter and 18.27% year over year, beating the upper end of its revenue forecast by about 5.2% and allowing the company to deliver the performance typically expected during the traditional peak season.

Advantech reported on October 7 that third-quarter consolidated revenue reached NT$32.44 billion (approx. US$1.02 billion), a record high as demand for edge AI, semiconductor equipment, and robotics stayed strong. September consolidated revenue rose to NT$10.749 billion, up 73.14% year-over-year, while cumulative revenue for the first nine months of 2026 increased to NT$78.952 billion from NT$52.961 billion a year earlier.

Semiconductor test interface maker WinWay Technology said AI and high-performance computing (HPC) demand, along with production-efficiency gains on some lines, drove shipments higher in 2026. The company surpassed NT$10 billion (approx. US$314.7 million) in cumulative revenue in September, a milestone it originally expected to hit in 2027, and reported record third-quarter sales as the semiconductor supply chain continued to benefit from strong AI-related spending.

SpaceX is reportedly seeking US$40 billion in financing to purchase Nvidia chips, a massive capital commitment driven by surging artificial intelligence (AI) computing demand. Viewed alongside Elon Musk's Terafab semiconductor manufacturing initiative, however, the move suggests a dual-track strategy: relying on Nvidia for near-term computing capacity while gradually reducing dependence on external suppliers over the longer term.

Infineon has finished acquiring India-based C2i Semiconductors, a maker of software-defined multiphase controllers and smart power stages used in AI data centers. The acquisition indicates the growing importance of power stability and efficiency as the number of energy-hungry AI processors rises.

India-based engineering firm Sasken Technologies plans to license company-owned semiconductor interconnect intellectual property (IP) to multiple customers, marking an effort to move beyond customer-specific chip engineering work toward a reusable IP business model.

Malaysia's plan to climb the semiconductor value chain is colliding with two constraints: a shortage of land in Penang, the heart of its chip industry, and mounting pressure to pick sides in the US-China technology contest. Off Penang's coast, developers are reclaiming a 2,300-acre Silicon Island from the sea to give the cluster room to expand, even as Washington presses Kuala Lumpur to police advanced chip flows and Chinese-linked firms are setting up in the state to reduce their exposure to scrutiny of China-origin goods.

Marvell's higher long-term target for custom AI chip revenue suggests the market for artificial intelligence hardware is widening beyond accelerators, into management chips, storage, memory, and networking. The shift could benefit Taiwanese design firms, but it also raises the bar for custom silicon expertise and customer access.

Huawei rotating chairman Eric Xu said he believes the company's Ascend AI chips have surpassed Nvidia in market share in China, but that domestic demand still cannot be fully met. He added that Huawei has no plan to broadly expand overseas at this stage, according to reports from Ijiwei, Sina Finance, and other sources.

Taiwanese IC design giant Realtek reported September 2026 consolidated revenue of NT$12.378 billion (US$388.5 million), down 0.7% month-over-month and up 23.35% year-over-year. Third-quarter revenue reached NT$37.732 billion, up 0.48% quarter-over-quarter and 27.94% year-over-year, while cumulative revenue for the first nine months of 2026 came to NT$111.708 billion, up 15.85% year-over-year.

Huawei executive director and consumer business group chairman Richard Yu said at a recent media roundtable that the company has worked with China's supply chain since US export controls were imposed in 2019 to build an independent supply system spanning chips, operating systems, basic software, and materials. He said Huawei has now essentially broken free from dependence on US technology.

Elon Musk has moved to quell uncertainty over Intel's position in the Terafab chip project, stressing that his companies will build and operate the planned semiconductor complex themselves while limiting any potential TSMC involvement to subleasing part of the site.

Before generative AI rose and Nvidia GPUs came to dominate the market, the central processing unit (CPU) was the main compute engine for servers. Now, as the personal AI agent boom gathers pace, CPUs are drawing renewed attention for their role in AI — a trend that is giving Advanced Micro Devices (AMD) powerful growth momentum.

Nvidia's RTX Spark superchip for Windows PCs moved from stage demos to order books on October 7, as pre-orders opened for Windows laptops built to rival Apple's unified-memory Macs at running large AI models locally. Reuters said cracking the Windows PC market could help Nvidia go after one of the last major markets dominated by Intel and AMD.

Samsung Electronics has become the first South Korean company to clear KRW100 trillion in operating profit in a single quarter, a milestone that shows how heavily the company now leans on a memory market where, by analysts' reckoning, suppliers hold pricing power. The open questions are whether high-bandwidth memory (HBM) contracts can extend that run into 2027, and how long the company's consumer-device arm can absorb the rising component costs that the same boom creates.