Taiwan's semiconductor industry and the military implications of a potential cross-strait conflict dominated the aerospace and defense track on the second day of the Plug and Play Silicon Valley May Summit, as investors and defense technology executives described a geopolitical environment that is fundamentally reshaping where capital flows and why.

On May 20, 2026, DIGITIMES held its Tech Forum, where senior analyst Jim Hsiao delivered a presentation, "AI server market outlook and trends amid the explosion of agent applications." In it, Hsiao explained that demand is rising sharply not only for x86 server CPUs from Intel and AMD, but also for Arm-based CPUs, which are expected to see explosive growth in 2026.

Nvidia has positioned its specialized LPX accelerator as a niche product designed specifically for low-latency, high-speed token generation rather than broad-market enterprise workloads. The hardware targets a narrow segment of service providers that operate premium, high-velocity token applications for select customer bases. While the architecture excels at rapid text decoding, its broader market adoption remains constrained by structural limitations in throughput and memory capacity.

Nvidia's H200 was a major focus after the Trump-Xi meeting, but hopes for sales into China have faded after US President Donald Trump's latest remarks. Trump said Chinese President Xi Jinping is firmly committed to developing domestic AI chips, leaving little room for common ground on the issue.

AMD said on May 21 that it plans to invest more than US$10 billion across Taiwan's semiconductor ecosystem to deepen strategic partnerships and expand advanced packaging capacity for next-generation artificial intelligence (AI) infrastructure. The initiative aims to strengthen manufacturing capabilities, accelerate the deployment of AI systems, and support the growing demand for high-performance computing hardware.

Nvidia has unveiled a strategic expansion of its silicon portfolio, detailing a four-pronged commercial approach for its new Vera central processing unit. The chipmaker expects the processor to generate US$20 billion in standalone revenue, signaling a significant evolution in its hardware distribution strategy and a broader play for infrastructure dominance in the emerging agentic artificial intelligence market.

Nvidia has detailed a significant structural shift in its data center revenue reporting, introducing a new market segmentation that separates traditional hyperscale cloud providers from a rapidly expanding sector named ACIE, which stands for artificial intelligence clouds, industrial, and enterprise. While hyperscale platforms currently lead initial deployment, management projects that the ACIE segment will eventually outgrow them, owing to the massive scale of the global corporate economy.

Demand for power components is surging as AI servers adopt high-voltage direct current power delivery and advanced cooling technologies. This shift could increase component density and design complexity across the data center supply chain. Taiwanese power semiconductor makers are moving into cooling and power-management MOS technologies, as well as higher-spec upgrades.

During a keynote discussion session at Tech Forum 2026, DIGITIMES senior reporter Monica Chen and semiconductor analyst Andrew Lu shared their views on current AI server technology transformations, concerns over capital expenditure bubbles among global cloud giants, changes in application-specific integrated circuits (ASICs) and chip architectures, and TSMC's global expansion strategy.

Nvidia introduced a new segment reporting framework on its first quarter of fiscal 2027 call, splitting data center revenue into Hyperscale and ACIE (AI Clouds, Industrial, Enterprise) and breaking out Edge Computing as a separate platform.

Nvidia used its first quarter of fiscal year 2027 earnings call on May 20, 2026 to lay out a three-tier silicon cadence that should make any rival roadmap look thin: a Blackwell ramp the company calls the fastest in its history, the first production silicon of Vera Rubin in the second half of this year, and a brand-new Arm CPU, Vera, that opens a US$200 billion TAM Nvidia has never touched.

Nvidia continues to exclude Chinese data center compute revenue from its outlook, with CFO Colette Kress citing uncertainty around whether H200 imports will be allowed into the country despite recent US export license approvals.

As AI computing demand continues to grow exponentially, pressure to upgrade data center network architectures is intensifying. Co-packaged optics (CPO) is moving from concept to real-world deployment, and Foxconn Industrial Internet (FII) is quietly building a strong market position by leveraging its existing strengths in CPO all-optical switches as well as AI servers.

May 20, Nvidia reported first-quarter fiscal 2027 revenue of US$81.6 billion, up 85% year-over-year and 20% sequentially. GAAP diluted EPS reached US$2.39, up 214% year-over-year, and non-GAAP EPS was US$1.87, up 140% year-over-year. Both topped consensus estimates, with revenue near US$79 billion and non-GAAP EPS of US$1.77, S&P Global Market Intelligence reported.

With less than an hour remaining before a planned strike was set to begin, Samsung Electronics and its labor union reached a tentative agreement late on May 20, narrowly avoiding what industry observers estimated could have triggered supply-chain disruptions worth more than KRW100 trillion (approx. US$66.8 billion).

Alibaba Cloud has unveiled the Zhenwu M890, a new proprietary AI training and inference chip developed by its semiconductor subsidiary T-Head Semiconductor, as the Chinese cloud group accelerates its push into full-stack AI infrastructure for the agentic AI era.

The global semiconductor industry has been focused on a memory supercycle, with some forecasts suggesting the upturn could last until 2030. However, Kye-hyun Kyung, former head of Samsung Electronics' Device Solutions (DS) division and currently a standing advisor to Samsung, has cautioned against overly optimistic sentiment in the memory market.

Recent media reports say Google is set to team up with Blackstone to form a new cloud computing leasing company, with Blackstone as the main shareholder. The new venture aims to build about 500 MW of computing capacity by 2027, most of it using Google's Tensor Processing Units (TPUs), a move industry watchers say supports Google ASIC partners such as MediaTek, Broadcom, and TSMC.

As artificial intelligence reshapes the global economy, demand for data centers is rapidly redrawing the export map of nations.

US President Donald Trump's trip to China with 17 business leaders thrust Tesla CEO Elon Musk and Nvidia CEO Jensen Huang back into the spotlight — as Beijing's position on Nvidia's H200 chips and China's broader AI supply chain continue to reshape the market narrative.

Intel is urging notebook and PC customers to adopt its most advanced chip manufacturing technology as surging demand for artificial intelligence computing continues to strain global CPU supply, according to industry sources.

Rising manufacturing costs at upstream foundries and back-end packaging and testing are pushing power management IC (PMIC) makers toward price increases, with global implications for device makers and automotive and data-center operators as firms weigh whether to accept higher PMIC prices or shift strategies amid ongoing worldwide capacity constraints and node migration pressures.

ITE Tech's stable first-quarter results and planned price increases from the second half of 2026 signal potential cost pass-through across global PC, notebook, and embedded markets, affecting device makers and buyers worldwide as memory and component costs climb and inventory strategies shift in response to anticipated further price rises.