Sandisk sees demand for roughly 60 to 90 gigabytes of NAND flash per CPU in certain AI inference and agent-orchestration applications, as customers rethink how they use different memory and storage technologies across their AI systems.
Passive component manufacturer Yageo said robust AI demand is driving shipments of multilayer ceramic capacitors (MLCCs) and tantalum capacitors, pushing AI-related applications to nearly 20% of its total revenue in the third quarter of 2026, up from 16% in the preceding quarter.
Intel's growing server business is strengthening its operating recovery, but a factory disruption and a prolonged PC downturn threaten its ability to convert demand into near-term sales. Its third-quarter results show why the turnaround now depends as much on supplying data centers as on reviving the client business that has long anchored the company.
Oracle will offer cloud services powered by Cerebras' AI computing systems in its cloud data centers starting in 2027, giving the wafer-scale chipmaker another major cloud customer. Although the companies did not disclose Oracle's planned Cerebras hardware purchases, the deal highlights the shift toward diversified AI compute resources as inference demand accelerates.
As AI chip packages continue to grow in size, panel-level packaging equipment is gradually advancing toward customer validation and volume production.
Sandisk says AI infrastructure customers are returning to secure more NAND flash after underestimating their storage requirements, even as its pipeline of prospective long-term supply agreements exceeds planned output through the end of the decade.
Amazon Web Services (AWS) saw its revenue grow 39.3% year-over-year in the third quarter of 2026. That was the cloud unit's fastest pace in nearly five years and its sixth straight quarter of acceleration. Segment sales reached US$46.0 billion, which puts AWS at a US$184 billion annualized run rate. Operating income rose 55% to US$17.8 billion.
AI chipmakers' rising memory performance requirements are accelerating the shift from sixth-generation high-bandwidth memory (HBM4) to HBM4E, according to analysis of the South Korean supply chain. Major AI semiconductor customers are already pushing HBM4E development, potentially shortening HBM4's mass-production cycle, while suppliers' development and manufacturing capabilities, thermal management, and customer validation remain decisive.
Strong demand for AI applications and cloud services is driving investment in high-performance computing chips and AI servers, prompting Taiwan manufacturers to expand capacity and borrow from banks while retaining cash. At the same time, banks' property-related loan balances continue to decline as credit resources flow into production and the real economy.
Semiconductor epitaxy maker Episil-Precision said a localized fire broke out on October 28 at its plant on Innovation First Road in Hsinchu Science Park, but caused no injuries or trapped personnel. The affected facility is one of the company's five production sites and accounts for less than 10% of its total output value and production capacity.
STMicroelectronics (ST) reported third-quarter 2026 revenue of US$3.76 billion, above the midpoint of its guidance, as AI data center and low Earth orbit (LEO) satellite demand lifted communication equipment and computer peripheral sales by 64% year-over-year. Gross margin reached 37.3%, up 410 basis points from a year earlier, and non-US GAAP earnings per share came in at US$0.37. Free cash flow was US$351 million.
More coverage
