At the 2026 World Economic Forum in Davos, the focus of discussions among technology leaders on AI has shifted from the dazzling competition over model parameters or breakthroughs in algorithms.

"2026 marks the year of AI's practical deployment," Jeetu Patel, President and Chief Product Officer of Cisco, remarked during the Davos Forum. However, he identified three major obstacles: infrastructure limitations, trust issues, and data-related challenges. Specifically, he noted, "Globally, there's a severe shortage of power, computing capacity, and network bandwidth to keep pace with AI's growing needs."

In 2026, global AI development faced its most concrete hurdle—a scarcity of power supply. Industry experts pointed out that AI workloads were already consuming tens of gigawatts of electricity worldwide, with projections suggesting this could surge to hundreds of gigawatts by the century's end, nearing the limits of current power infrastructure.

"2026 is the year when power shortages start to seriously impede AI's progress," cautioned Varun Sivaram, Founder and CEO of Emerald AI, at the Davos gathering. He disclosed that the U.S. plans to construct 50 gigawatts of data center capacity over the next three years, yet only 25 gigawatts can be integrated into the power grid.

China, on the other hand, is anticipated to have 400 gigawatts of surplus capacity available for AI applications by 2030. The electricity demand for AI is skyrocketing. Unlike traditional data centers, AI data centers necessitate continuous high-power operation, posing unprecedented challenges to the stability and capacity of power grids.

Cisco's Patel observed that power constraints are reshaping the architecture of AI infrastructure, leading to an increasing interconnection of multiple data centers to function as a single virtual cluster. This distributed architecture not only addresses power shortages but also enhances resource utilization.

As data center construction drives up electricity bills for U.S. households, tech giants are recognizing that technological prowess is no longer the only barrier—gaining social acceptance is equally vital.

The cancellation of Microsoft's Caledonia data center project in Wisconsin due to local opposition marked a pivotal moment. This setback prompted Microsoft to reevaluate its infrastructure strategy.

Microsoft subsequently introduced a community-centric framework, pledging rate structures that avoid shifting costs onto residential users and funding the new power capacity and grid upgrades required for its data centers.

The company also committed to paying full local property taxes, supporting public services such as schools, hospitals, parks, and libraries, and setting a target to reduce water consumption intensity in data centers by 40% by 2030.

"The most pivotal decisions are often made at the grassroots level," remarked Brad Smith, Microsoft's Vice Chair and President, "highlighting the profound impact of infrastructure expansion on local communities."

With the proliferation of AI agents—industry research firm IDC projects over 1 billion AI agents will be operational in the global economy by 2029—trust and transparency have become even more paramount.

As AI development enters a new phase, enterprise demands are evolving from mere functionality to user-friendliness, and tech companies themselves are undergoing role transformations. In this new era, simply providing computing power or models is no longer adequate.

Google has raised its annual capital expenditure plan to roughly $85 billion, with projections indicating continued increases in 2026. Its cloud business sales surged nearly 32% year-over-year, surpassing growth expectations. Meanwhile, Google's AI infrastructure head informed employees that the company must double its computing capacity every six months to meet the burgeoning demands for AI services.

CEO Pichai described 2026 as an "exceptionally intense year," citing heightened AI competition and the pressure to fulfill cloud and computing needs.

NVIDIA recently appointed its first Chief Marketing Officer, who reports directly to Jensen Huang. This move signifies more than just a marketing strategy—it's a signal of strategic transformation. NVIDIA is transitioning from a GPU supplier to a platform and infrastructure provider, necessitating a clear explanation to global businesses, governments, and industries on how to deploy, utilize, and sustain computing power in the long run.

References:

https://finance.sina.com.cn/hy/hyjz/2026-01-19/doc-inhhvsft7967847.shtml?cre=tianyi&mod=pchp&loc=2&r=0&rfunc=9&tj=cxvertical_pc_hp&tr=12

https://www.cnbctv18.com/technology/wef-davos-2026-ai-adoption-power-security-enterprise-readiness-tech-ceos-19824094.htm/amp

https://ai.zhiding.cn/2026/0116/3177073.shtml

https://m.zhitongcaijing.com/article/share.html?content_id=1321270