Anduril Industries said it will spend US$3.7 billion to build Arsenal-2, a next-generation software-defined shipyard at Sparrows Point in Baltimore County, Maryland. The US Navy is backing the project with a contract worth up to US$2.9 billion tied to production results, bringing the total commitment to US$6.6 billion as Washington seeks to expand the submarine industrial base.
Apple and LG Electronics are reportedly pairing up to create a line of smart home devices, though the devices will carry the LG brand name. The two companies are expanding their reach into the smart home market as both face headwinds in their home product segments.
SpaceX is seeking roughly US$40 billion in financing to fund purchases of Nvidia AI chips, in a deal led by Apollo Global Management that could become one of the largest examples yet of compute hardware being financed as long-duration infrastructure.
Meta and enterprise AI startup Sierra are leading Walmart, Stripe, Shopify, Genesys, Instinct and Rocket in developing an open standard for personal AI agents, aiming to define how software acting for consumers can authenticate, obtain permission and interact with businesses.
Schneider Electric's roughly US$22.6 billion acquisition of US industrial software developer PTC is more than a bid to expand its computer-aided design (CAD) and product lifecycle management (PLM) software lineup. It signals that industrial AI competition is shifting from point automation and smart manufacturing to a platform race spanning product design, engineering, deployment, energy management, operations, and maintenance.
Huawei and Qualcomm announced on Oct. 5 that they had signed a multi-year patent licensing agreement covering technologies including 5G, computing, AI and networking.
AMD CEO Lisa Su moved from supply-chain talks in Taiwan to South Korea this week, broadening the chipmaker's Asia push from efforts to secure more CPU, GPU and advanced-wafer capacity to building an open computing ecosystem around its processors, Korean AI accelerators and software.
Rising memory prices are already weighing on 2026 smartphone and notebook shipments, and sustained AI infrastructure investments are expected to keep quotes climbing into 2027. To counter softer buying interest as retail prices rise, hardware vendors like Samsung are pivoting to fresh strategies. Alongside cost controls and broader product line-ups, companies are increasingly prioritizing software differentiation, AI integration, post-purchase services, and aggressive promotion of previous-generation (N-1) models.
As AI agents integrate into enterprise operations, companies scaling deployment face rising challenges around runaway costs, permission controls, and cybersecurity governance. CloudMile founder and chairman Spencer Liu noted that many firms using AI still lack clarity on cost allocation or what data agents have accessed.
A string of AI server smuggling prosecutions is showing that following US export-control rules to the letter has not stopped advanced Nvidia hardware from reaching China. The weak point lies less in the chips themselves than in the multi-layered server supply chain that carries them, where each participant checks its own slice, and no one owns the final destination.
The global artificial intelligence landscape is witnessing an accelerating divide. While Silicon Valley grapples with escalating safety debates and regulatory friction, Chinese tech developers are rapidly deploying lightweight models and domestic chips. Simultaneously, European players are stepping up to claim AI sovereignty through open-weight models designed for critical enterprise infrastructure.
Chinese AI company Z.ai is expanding its overseas commercialisation through Amazon Bedrock, with AWS adding GLM-5.3 on October 6 and sharing revenue with Z.ai based on model usage.
SpaceX launched its Starship super heavy-lift rocket on September 28, successfully reaching orbit for the first time and deploying 26 Starlink V3 low Earth orbit (LEO) satellites. Each Starship mission can carry up to approximately 60 Starlink V3 satellites. Although neither stage was recovered, both splashed down at sea and the satellite deployment was successful.
Taiwan's tech industry has grown rapidly in recent years on the back of strengths in semiconductors, servers and the data center supply chain, lifting overall GDP. But companies that keep chasing hardware specs or wait for technology to become perfect before deploying AI will face an endless cycle of catch-up pressure, according to former minister of science and technology Liang-Gee Chen.
Hwacom Systems has shifted its semiconductor-focused business to subsidiary SureWin, targeting growing demand from chipmakers for AI deployment, data governance, and cybersecurity, including post-quantum cryptography (PQC).
Chinese AI startup DeepSeek is poised to raise more than CNY80 billion (approx. US$11.9 billion) in its latest funding round, as it builds capital ahead of a potential domestic initial public offering (IPO).
India is joining the top tier of Asia Pacific data center markets with a stock of facilities designed for a fraction of the power AI hardware draws. That gap is turning the country's build-out into a race for liquid cooling, power equipment, and high-density designs, not just floor space.
Huawei is backing near-packaged optics (NPO) rather than co-packaged optics (CPO) for its next-generation AI infrastructure, arguing that NPO offers a better balance of cost, manufacturability, reliability and serviceability as optical interconnects move deeper into large-scale AI clusters.
The global race to build data centers for artificial intelligence is entering a more difficult phase. Demand for computing capacity remains strong, but new projects increasingly face limits on land, electricity, water, and local acceptance, turning infrastructure expansion into a broader political and social contest.
The traditional market divide between the third-generation semiconductors gallium nitride (GaN) and silicon carbide (SiC) is seeing greater overlap in 2026, with SiC increasingly encroaching on GaN's territory.
Marvell Technology is betting that the next phase of AI infrastructure growth will require substantially more networking and optical connectivity, raising its fiscal 2028 revenue outlook to approximately US$20 billion and setting a fiscal 2031 target of US$70 billion to US$90 billion. The company's strategy centers on extending existing hyperscaler relationships across switches, optics, and custom chips surrounding AI processors.
AI server supply chains are accelerating their shift toward regional manufacturing. However, the labor access, wage costs, and capacity expansion constraints of US manufacturing are forcing manufacturers to raise automation levels to balance delivery speed and production efficiency. That has pushed Foxconn and Nvidia to jointly invest in precision assembly automation research, with an eye toward reducing manual handling.
Catcher Technology said its existing AI products continued to expand shipments as the year-end period approached, while medical and drone businesses were entering mass production. The casing maker also said additional AI server projects are scheduled to begin mass production from the first quarter of 2027, with AI PC-related products expected to follow in the second quarter of 2027.
Transcend reported a sharp recovery in September 2026 revenue as shipments for industrial PC and networking customer projects picked up. The memory module maker said monthly consolidated revenue reached NT$6.04 billion (approx. US$190 million), up 24.8% from August and 290% from a year earlier.
