When the agenda for this year's OCP APAC Summit landed, seasoned observers noticed something unusual. TSMC, Applied Materials, Advantest, and ASE are sharing the stage with Microsoft, Nvidia, and Google.
For three years, the constraint on artificial intelligence was how much compute the hyperscalers wanted to buy. The June-quarter filings of seven companies say the constraint has moved. It is now about who sets the price — and on that question the past five quarters have already taught a lesson from which most readers drew the wrong conclusion.
Delta Electronics expects capital spending to rise to roughly NT$70 billion(US$2.17 billion) in 2026 from NT$46.1 billion(US$1.43 billion) in 2025, driven by growing demand from AI data centers and energy infrastructure projects.
US transportation authorities have allowed Amazon's Zoox to operate commercially on a limited basis, making it the first paid robotaxi service approved in the country. This follows efforts by Trump administration officials to relax certain rules in an effort to encourage development in autonomous driving technology.
Amazon disclosed a US$496 billion AWS backlog growing at triple-digit rates in the second quarter, and CEO Andy Jassy used the call to lay out unusually specific unit economics defending the capital behind it.
LG Electronics reported a strong second quarter, supported by record sales and one-time gains, while accelerating a shift toward robotics and artificial intelligence infrastructure. The company is increasingly framing its future around physical AI, even as it continues to rely on premium appliances, televisions, and vehicle electronics to support near-term earnings.
The rapid expansion of AI infrastructure is reshaping one of the semiconductor industry's most overlooked component markets. Soaring demand for high-end multilayer ceramic capacitors (MLCCs) is prompting suppliers to reallocate production capacity and driving a new round of price increases that extend beyond AI-specific products.
Delta Electronics expects capital spending to rise to roughly NT$70 billion in 2026 from NT$46.1 billion in 2025, driven by growing demand from AI data centers and energy infrastructure projects.
ams OSRAM is developing Micro LED-based optical data links for AI data centres, positioning the technology as a high-density, scalable alternative to copper interconnects. To support adoption, the company has built a high-precision simulation platform that evaluates Micro LED optical links under real AI workloads.
Microsoft CEO Satya Nadella laid out how the company is scaling its AI infrastructure and platform to keep pace with demand. He outlined these plans during an earnings call on July 29, when he also announced record revenue above US$331 billion for fiscal year 2026 (ending on June 30), with Microsoft Cloud earnings surpassing US$214 billion and Azure earnings surpassing US$100 billion.
Microsoft reported stronger-than-expected fiscal fourth-quarter results, driven by accelerating Azure cloud growth and rising adoption of its AI products, reinforcing investor confidence that the company's heavy investments in AI are generating commercial returns. Robust guidance for the current quarter and continued enterprise demand also eased concerns over spending on AI infrastructure.
Arm Holdings is preparing its first self-designed data center CPU, the Arm AGI CPU, for shipments by the end of calendar 2026, and expects silicon sales to become a separately reported revenue category in fiscal 2028. The change would give Arm a third reported revenue line alongside licensing and royalties as it expands from supplying processor designs to selling complete chips.
Global automakers are facing slower market growth, the rapid rise of Chinese carmakers, and sustained EV investment, with Volkswagen, BMW, and Mercedes-Benz under strain in recent years. While German automakers still lead the global auto industry, revenue and profitability have come under broad pressure over the past two years as competition shifts from volume growth to product value and operating efficiency.
A Taiwan startup says it has found a way to hide advanced eye-tracking technology inside what looks like an ordinary pair of glasses — using just one camera and a single LED light source.
Nvidia CEO Jensen Huang once again voiced his support for open-weight and open-source AI, a form of model that has caught official attention in Washington after the release of the powerful Kimi K3 model from Chinese startup Moonshot AI. His comments added another voice to the ongoing debate over closed-source versus open-source approaches as the US considers restrictions on Chinese AI.
NXP Semiconductors' physical AI strategy extends beyond vehicles, factories and robots. The same control, connectivity and real-time processing capabilities are also giving the company a larger role in AI data center infrastructure, where it is targeting the systems that manage networking, cooling, security and power around increasingly complex accelerator platforms.
