Speaking at the Silicon Photonics Global Summit ahead of Semicon Taiwan 2026, Marvell Technology CTO Radha Nagarajan highlighted that the core focus for upgrading connectivity in the AI data center era has shifted firmly to power consumption. He noted that the acceleration of co-packaged optics (CPO) adoption is primarily driven by two critical bottlenecks: power and unit density.

As AI packages grow larger, the next substrate battleground is rapidly extending from traditional organic materials to glass. Samsung Electro-Mechanics (SEMCO) will make its first appearance at SEMICON Taiwan 2026, with advanced packaging and glass substrate expert Gang Duan traveling to Taiwan to unveil the company's roadmap for next-generation advanced packaging substrates aimed at AI infrastructure.

Anthropic has reportedly signed a cloud-computing contract worth US$35 billion with Lambda, an Nvidia-backed cloud provider, under which Nvidia itself holds the data center lease, The Wall Street Journal reported, citing people familiar with the deal. The site is being developed by Hut 8, a bitcoin miner turned data-center builder, in Nueces County, Texas, and Nvidia signed an agreement with Hut 8 a few weeks ago to secure capacity, according to the Journal.

South Korean chip startup XCENA is pushing CXL beyond memory expansion and toward a new layer of server computing, unveiling measured results for its MX1 computational memory device at Hot Chips 2026 while expanding its work with Samsung Electronics on a rack-scale system.

India has unveiled the next phase of its semiconductor policy, a move that could reshape global chip supply chains, investment flows, and talent markets. The expanded program aims to deepen local design and manufacturing capacity, while giving multinational companies and developers a larger role in one of technology's most strategic sectors.

SEMICON Taiwan 2026 is set to open soon as Semiconductor Equipment and Materials International said AI is pushing a new wave of technology upgrades across the global chip industry. At an August 31 pre-event briefing, the group said the changes are accelerating work in AI computing, advanced process technology, heterogeneous integration, and quantum computing.

Nvidia's US$3.5 billion investment in MediaTek's overseas convertible bonds marks more than a financial bet. It underscores how the chip leader is trying to anchor its AI infrastructure strategy as cloud giants and model developers expand their use of custom silicon and rack-scale systems.

Nvidia and MediaTek are widening a global AI partnership that could reshape data center, edge, and PC hardware markets. MediaTek's US$3.9 billion overseas convertible bond sale, largely bought by Nvidia, signals a deeper strategic tie that may affect supply chains, customer choices, and competition worldwide.

As trillion-parameter large language models (LLMs) and agentic AI become mainstream, the AI infrastructure bottleneck is shifting from raw compute power toward system-level integration across compute, memory, interconnects, and thermal management. To bypass traditional hardware constraints, Nvidia announced NVHBM, its next-generation custom HBM technology, and integrated it into the NVLink Fusion ecosystem.

Silergy said at a recent earnings call that most of its applications, excluding its computing product line, posted quarter-over-quarter growth of more than 20% in the second quarter of 2026. Automotive revenue rose about 50% year-over-year, industrial revenue increased 49.7%, and data center-related applications grew nearly 90% year-over-year.

MediaTek has completed pricing on US$3.9 billion of overseas convertible bonds, more than NT$120 billion and what the company says is the largest overseas CB ever issued in Taiwan's capital market. Nvidia subscribed for US$3.5 billion of it — roughly 90% of the offering.

Marvell said on its fiscal second-quarter of 2027 earnings call that customers are actively planning to deploy scale-up optical solutions as early as 2027, with demand for near-package optics (NPO) now revised higher. The company said copper cabling is increasingly constrained by distance and bandwidth limits as computing systems grow larger, pushing customers to raise the share of optical transmission in their designs.

Nvidia reported second-quarter revenue for fiscal year 2027 of US$96.22 billion, a 106% year-over-year increase, driven by a 117% surge in data center revenue to US$89 billion. With third-quarter revenue of fiscal year 2027 projected to hit US$108 billion and full-year fiscal year 2028 revenue expected to grow around 70%, Nvidia revealed that current supply capabilities meet only about 70% of customer demand. This indicates that the primary constraint on Nvidia's sustained hyper-growth has officially shifted to the supply side.

Chip design company Agentrys announced last week that it raised US$24.5 million in funding from backers including MediaTek. It is one among an emerging crop of startups using agentic tools to power the design of semiconductors, a step in the chip process that still involves considerable amounts of time and labor.

MetaX posted its first half-year profit since listing, with rising GPU shipments and the mass production of its C600 accelerator strengthening its commercial position while China's domestic AI chip sector increasingly splits between profitable vendors and peers still absorbing heavy R&D costs.

The AI infrastructure race has spent the past few years fixated on high-bandwidth memory as one of the most indispensable building blocks of accelerated compute. Yet as model sizes, context windows and inference workloads continue to expand, the bottleneck is increasingly shifting from how much memory can be stacked toward how efficiently data can move between memory and compute.

The four major US cloud service providers (CSPs) have already rewritten their 2026 capital expenditure records, and ASIC vendors are increasingly confident that demand will remain strong through 2028 as customer spending and supply-chain visibility both extend farther out.

Apple on August 25 unveiled the M6 and M5 Ultra chips without holding a launch event, with the M6 becoming the company's first processor built on TSMC's 2nm process and debuting in the new Mac mini. But industry watchers say the M6's short-term significance lies mainly in showcasing 2nm design technology, not in delivering immediate commercial gains.

In an effort to accelerate its custom silicon initiative and curb reliance on external hardware suppliers, artificial intelligence developer Anthropic held discussions to acquire AI chip startup MatX for roughly US$7 billion, according to Reuters. While the acquisition talks have since evolved into potential partnership discussions, the move underscores Anthropic's ambition to expand its proprietary hardware capabilities.

Huawei's Ascend 910C is already sold out in China, but domestic chips still cannot fully replace Nvidia's H200s. The shortage matters beyond China because AI compute constraints, pricing, and supply chain shifts are reshaping how quickly companies worldwide can deploy advanced inference systems.

According to Bloomberg, Nvidia has formed a federal political action committee (PAC), expanding its Washington presence as governments debate artificial intelligence rules that could affect businesses, workers, and consumers worldwide. The move underscores how major technology firms are preparing for a longer policy battle over AI, competition, and market access.

Nvidia's latest earnings and forecast suggest AI demand remains far ahead of available supply, with implications for memory, power, cooling, and networking worldwide. The chipmaker's guidance points to continued price pressure and a broader buildout of infrastructure that will shape AI adoption across markets and industries.

Samsung Electronics' packaging technology lead, Lee Hee-seok, outlined the company's advanced semiconductor packaging roadmap at Advanced Semiconductor Packaging & Chiplet Show (ASPS) 2026. Beyond sharing target production timelines and specifications for fan-out panel-level packaging (FOPLP) and glass substrates, Lee emphasized AI's pivotal role in driving packaging R&D forward.

Surging AI chip demand continues to outstrip TSMC's advanced packaging capacity. At the same time, massive AI server shipments are driving up demand for CPUs, networking equipment, power management ICs (PMICs), and peripheral silicon, rapidly soaking up traditional packaging capacity. IC design insiders reveal that several outsourced semiconductor assembly and test (OSAT) providers have already warned clients that traditional packaging capacity could face a deficit exceeding 20% by 2027. New capacity cannot be added quickly enough, with the supply of wire bonding equipment emerging as the primary bottleneck.