Global AI growth is increasingly colliding with electricity limits, a shift that could slow data center buildouts and reshape infrastructure planning from the US to Asia. Delta Electronics chairman Ping Cheng said the bottleneck is already delaying projects, pushing operators toward self-owned power systems and off-grid microgrids.

Quanta Computer executive vice president and Quanta Cloud Technology (QCT) president Mike Yang said GPU applications are expanding across training, edge computing, and storage, as well as in the development of application-specific integrated circuits (ASICs). In response to future demand, the company is preparing for power needs and production capacity, planning to add three more plants in the US by the end of 2026.

Lightmatter has joined Nvidia's NVLink Fusion ecosystem, a move that could accelerate the rollout of high-performance optical links for AI infrastructure worldwide. The collaboration is aimed at easing data-center bottlenecks, improving bandwidth, and giving global customers more options for building energy-efficient AI systems at scale.

Strong demand for AI server cooling is extending visibility for the industry, with Asia Vital Components (AVC) and Auras Technology both forecasting sustained growth through 2029. Nvidia is driving a shift from air cooling to liquid cooling in AI servers, boosting both revenue and profits for cooling module makers.

COMPUTEX drew major AI supply chain players to Taiwan, and industry sources said SK Group chairman Chey Tae-won was set for a private meeting with Foxconn chairman Young Liu, alongside his meetings with TSMC chairman C.C. Wei and Nvidia CEO Jensen Huang. The reported talks could signal broader cooperation on AI infrastructure.

As Computex 2026 opened in Taipei on June 2, Taiwanese President Ching-te Lai said major international tech companies are increasing investment in Taiwan, underscoring confidence in the country's industrial efficiency and democratic system. He said protecting peace and stability across the Taiwan Strait is the government's most responsible commitment to the global supply chain that relies on Taiwan's AI industry ecosystem.

Largan Precision's debut at Computex 2026 signals a broader push into optical communications that could matter for AI data center supply chains worldwide. The Taiwan company is seeking new growth beyond lenses, and its co-packaged optics efforts reflect rising industry interest in faster, denser, and more efficient connectivity.

Delta Electronics said at COMPUTEX 2026 that it has launched a prefabricated AI modular data center designed to speed deployment of AI infrastructure by cutting IT build time by about 60%. The move underscores how vendors are racing to support global demand for denser, faster, and more power-efficient AI facilities.

At Microsoft Build 2026 on June 2, Microsoft outlined how its Azure cloud and broader AI portfolio are being re-engineered for agentic AI — software agents that reason, retrieve knowledge, take actions, and run continuously rather than responding to one-off prompts. Spanning silicon, databases, runtimes, developer frameworks, and governance standards, the announcements describe a stack designed to enable organizations to build agents, deploy them in production, and keep them under control.

Ampak Technology unveiled plans to further pivot into AIoT and Edge AI integrated solutions at COMPUTEX 2026, announcing product launches including an AI system-on-module, 5G Reduced Capability (RedCap) offerings, and High Power Wi-Fi 6 drone communication modules. The wireless communication module designer said the moves respond to an industry shift from cloud data centers to endpoint and edge computing and aim to capture demand from enterprise and industrial applications in the US and other markets.

At COMPUTEX 2026, energy and data center executives warned that the industry shift from AI training to inference and agentic AI has driven a sharp rise in electricity demand and tightened requirements for power reliability and local supply. The discussions centered on Taiwan, where executives and energy suppliers said on-site power generation and energy storage systems have become essential for balancing volatile loads and ensuring uninterrupted operations.

Nvidia CEO Jensen Huang's frequent appearances with global tech chairmen and CEOs signal more than AI hype. They point to a redistribution of influence across the next generation of AI infrastructure.

As artificial intelligence (AI) infrastructure grows larger and more complex, the challenge is no longer simply building data centers. It is building them fast enough.

As Europe and the US fall short of expectations for the automotive electrical/electronic architecture (E/EA) transition, traditional tier-1 suppliers are accelerating diversification efforts to offset slowing automotive growth. Among them, France-based Valeo, one of the world's top-15 automotive parts suppliers, is leveraging its automotive expertise to expand into faster-growing sectors including AI data center infrastructure, defense, robotics and small-mobility solutions.

Nvidia CEO Jensen Huang said the computing industry is entering an era of "agentic computing," in which data centers, personal computers, autonomous vehicles, humanoid robots, and satellite systems will share a common AI architecture. He framed the shift as a broad reworking of how future devices will operate.

Ennoconn, a Taiwan-based industrial PC (IPC) vendor, said on June 1 at a forum in Taipei that its physical AI business is expected to exceed NT$10 billion (US$318.93 million) in 2026, as it deepens ties with Austria-based smart IoT solutions supplier Kontron and expands commercialization. The company said the partnership could generate an additional NT$10 billion in synergies by 2028.

On the first day of Computex 2026 in Taipei, Marvell Technology CEO Matt Murphy made a pointed argument to an audience of industry insiders: the next major bottleneck in AI infrastructure is not compute or memory; it is connectivity.

Power Integrations (PI) has announced two ultra-thin, compact auxiliary power supply (PSU) reference designs built for 800VDC AI data centers. PI principal training engineer Jason Yan said the ultra-compact solution, designed for Nvidia Kyber liquid-cooled blade-rack systems, uses a highly integrated GaN design that saves 30% of PCB space and reduces component count by 30% compared with conventional silicon carbide (SiC) solutions.

Foxconn Technology Group used COMPUTEX 2026 to present a broader AI infrastructure push, signaling a move beyond AI server supply and into token factory services. The company highlighted its "3+3+3" transformation strategy spanning manufacturing, platform applications, robotics, healthcare, and space-related computing.

Intel is expanding its data center portfolio with new Xeon 6+ processors, Ethernet E835 networking products, and fresh details on its Crescent Island AI accelerator, positioning the CPU as a central control point for the next phase of AI infrastructure.

Alphabet said it will raise US$80 billion in equity to fund a rapid expansion of artificial intelligence infrastructure as demand for its services accelerates, signaling a more capital-intensive phase of growth that could reshape global cloud and chip supply chains. Berkshire Hathaway separately agreed to invest US$10 billion, underscoring investor conviction in the company's long-term AI build-out.

In 2026, a global compute shortage spanning chips, cloud services, servers, and data-center components is sweeping across the industry. The scarcity of compute and broad price hikes are running through the entire AI sector, pushing Nvidia's market value higher, lifting cloud revenue and profits at Amazon, Microsoft, and Google to record levels, and driving the valuations of AI startups OpenAI and Anthropic to nearly US$1 trillion.

Advantech on June 1 said it deepened its strategic partnership with Nvidia and launched an AI-native factory architecture centered on an agentic multi-agent system called AI Factory Brain, aimed at extending AI decision-making into real-time factory operations. The announcement described the move as a shift from hardware supply to integrated solutions, with Advantech using its own factories as validation sites for the new platform.