Broadcom has partnered with Apollo Global Management and Blackstone's Credit & Insurance business to launch the AI XPV Platform, a new financing vehicle backed by an initial US$35 billion commitment aimed at accelerating AI infrastructure deployment. According to The Wall Street Journal, the platform is designed to support more than 20GW of AI compute capacity by 2028 using Broadcom's chips and networking technologies, with customers including Anthropic and OpenAI.

EZConn's record May revenue points to continued demand in optical communications, with implications for customers, suppliers, and investors tracking global data center expansion. The company said deferred April orders helped shipments, and it expects capacity to stay tight, margins to improve, and full-year growth to remain in double digits.

As AI moves from large-model training to inference, more applications and business models are emerging across the supply chain. InWin is expanding from components into system assembly, while Y.S. Tech is ramping up production as it expects AI-related growth momentum to build in 2027.

Flytech announced a strategic shift from being primarily a point-of-sale hardware supplier toward a systems-provider model focused on payments, edge computing, and recurring services. At Computex 2026, the firm unveiled a full payment solution covering counter POS, self-service kiosks, and handheld payment terminals designed for retail and food service, and said the move reflects changing customer mixes and a push to embed computing power across store operations.

GEM Terminals reported consolidated revenue of about NT$428 million (US$13.6 million) in May 2026, up 71.83% year-over-year from NT$249 million, saying expanding global investment in AI infrastructure drove a surge in demand for specialty copper materials used in cooling applications. The firm linked the month's performance to rising cooling needs at servers and data centers as computing density increased, and described the result as evidence of strong long-term momentum for its materials transformation strategy.

Global hardware growth is facing an increasingly fragile and fragmented supply chain. At PCIM Europe 2026, software intelligence firm Luminovo's OEM Growth Lead, Inga Schwarz, made a compelling case for why AI is no longer enough to save hardware companies from costly operational challenges. The industry, she argued, must embrace a transition toward deep, native domain enterprise integration to build a unified "digital thread." With the fast-moving advancement of generative models set against modern supply chain complexity, Schwarz delivered a reality check for OEMs and EMS providers navigating the global market.

Computex Taipei 2026, held from June 2 to 5 under the theme "AI Together," drew more than 1,500 exhibitors from 33 countries and set a new record in scale. The show underscored a new AI industry reality: competition has moved far beyond standalone chip compute and into a systems-level battle spanning compute, connectivity, power, and cooling.

SuperAlloy reported that rising demand from global luxury and supercar manufacturers, alongside green procurement policies, drove higher sales of recycled aluminum in May, lifting group momentum into the first half of 2026. The company said consolidated revenue for May reached NT$635 million (US$20.14 million) — up 6.7% from April and 18.74% from the same month in 2025. Cumulative consolidated revenue for January through May 2026 totaled NT$3.103 billion, up 2.86% year over year, returning the company to positive growth for the period.

Computex 2026 has ended, with the spotlight again firmly on Nvidia CEO Jensen Huang. From his arrival in Taiwan on May 23, Huang spent two weeks meeting key industry figures, attending Nvidia developer events, GTC Taipei, and a Computex tour, and once again hosting his "trillion-dollar banquet."

As AI shifts from training to inference and agents, computing and power demand are growing exponentially, putting energy supply and load stability at the center of the industry's next challenge. At the recently concluded Computex 2026, Delta Electronics and Liteon Technology showcased power supply solutions as data centers grapple with "tokens per watt" efficiency metrics and sharp current peaks.

OpenAI has taken its clearest step yet toward becoming a publicly traded company, announcing that it has confidentially submitted a draft S-1 registration statement to the US Securities and Exchange Commission (SEC).

Amazon has reached a multiyear, multibillion-dollar agreement with Corning to supply optical fiber, cable, and connectivity equipment for its US data centers, a move that stands to ripple beyond North Carolina by reinforcing domestic manufacturing and the networks that support cloud services, artificial intelligence, and digital infrastructure worldwide.

Computex 2026 closed last week with physical AI among its central themes, and robots emerging as one of the clearest ways to demonstrate it. Yet, unlike CES, where robot makers competed to showcase their hardware, Computex presented a different picture: AI computing platforms, edge inference, physical AI architectures, and the ecosystems behind robots took center stage.

Genius Electronic Optical reported consolidated revenue of NT$1.629 billion (US$51.5 million) in May 2026, marking a month-on-month decline from NT$2.314 billion in April but a 30.24% increase from NT$1.251 billion in May 2025. For the period from January through May 2026, cumulative consolidated revenue reached NT$10.084 billion, exceeding NT$10 billion and representing a 19.67% rise from NT$8.426 billion in the same period of 2025.

Taiwan's leading server manufacturers reported strong May sales, underscoring how global AI infrastructure spending is reshaping demand for data center hardware. The gains point to continued momentum in the second and third quarters, even as notebook markets remain weak and broader personal computer demand remains subdued.

Molex is building out a dual-track strategy for AI interconnects, backing both copper and optical solutions as customers pursue different deployment paths, while expanding its Taiwan operations to support the region's AI hardware supply chain.

Digital camera demand has made a striking comeback as younger consumers, social media creators, and the streaming economy inject new momentum into a market many had considered obsolete. The resurgence is prompting suppliers to restart idle production lines and invest heavily in new tooling, even as uncertainty remains over how long the boom will last.

Nvidia and LG Group announced a wide-ranging partnership on June 7 covering AI factory infrastructure, home robotics, autonomous driving components, and sovereign AI model development, making it one of the broadest single-company collaborations Nvidia has announced during Jensen Huang's South Korea visit.

Nvidia and Doosan Group are widening their collaboration to develop physical AI, robotics, and AI factory infrastructure that could shape industrial automation worldwide. The effort spans robotics, heavy equipment, power systems, and advanced materials, highlighting how global AI growth is increasingly tied to manufacturing, energy, and data center supply chains.

Hitachi and Intel have agreed to work together on physical AI, advanced computing, and digital infrastructure, a move that could shape manufacturing, energy, and mobility systems used worldwide. The partnership targets efficiency, resilience, and faster industrial innovation, with potential implications for factories, power networks, and other critical operations globally.

Largan Technology is set to hold its annual shareholder meeting on June 9, and investors worldwide will be watching for signals on handset demand, manufacturing capacity, and its push into AI optical communications. The meeting could also indicate how quickly the lens maker can scale new technologies, and whether margin pressure is easing.

argan Technology reported consolidated May revenue of NT$4.593 billion on June 5, a 14% decline from April and a 43% increase from May 2025. First five-month revenue reached NT$25.495 billion, up 15% year on year. The company said it would hold an investor conference on June 9, where progress in common-package optics — CPO — is expected to be a primary focus.

ASRock Rack, a subsidiary of Pegatron Group's ASRock Inc., has won an order to supply 587 GPU servers equipped with Nvidia B200 accelerators to Japan-based AI infrastructure provider Datasection Inc., the two firms announced. The systems are contracted for deployment at Datasection's AI data center near Bangkok, Thailand, and will serve as a dedicated high-performance computing platform for US companies.