Davos concludes: Three judgments reveal the true watershed moment for AI

From June 23 to 25, the 17th Annual Meeting of the New Champions of the World Economic Forum concluded in Dalian. Over 1,700 guests from more than 90 countries attended, with the theme summed up in two words—"Scaling Innovation." AI sessions were packed to capacity, with long queues at the doors.

From June 23 to 25, the 17th Annual Meeting of the New Champions of the World Economic Forum concluded in Dalian. Over 1,700 guests from more than 90 countries gathered under a four-word theme — "Innovating with Scale". AI sessions were packed to capacity, with long queues forming at the doors.

Bustle does not equal usefulness. I carefully reviewed the three-day agenda and guest speeches, and what is useful for enterprise digital managers is not those flashy exhibits, but three judgments.

Judgment 1: Technology diffusion is much slower than the technology itself — this is the real bottleneck

Xue Lan, Dean of Schwarzman College at Tsinghua University, used an analogy during the sub-forum "AI is everywhere, but not achieved overnight": just as roads must be built first, then gas stations, and then traffic rules established, the deployment and promotion of AI requires both hardware construction like data centers and software support like regulatory rules.

Zhu Min, a member of the World Economic Forum's Board of Trustees, put it more bluntly:

The birth of cutting-edge technologies such as quantum sensing and AI from laboratories may not be slow, but the real challenge lies in their diffusion and application. Although the technologies themselves are accelerating in iteration, their pace of development in practical scenarios depends on whether the foundations of each country are solid.

Translation: Having the best AI model doesn't mean you can actually use it. The speed of technology diffusion depends on infrastructure, governance rules, talent pool, and data quality—the progress of these "soft supporting factors" is the real bottleneck for enterprise AI.

The judgment of FedEx China President Poh-Yian Koh is even more heart-wrenching:

Most AI projects lack full-process transformation, resulting in poor implementation outcomes. In the current era of rapid AI development, data quality, governance capabilities, and the depth of scenario-specific application determine the speed and breadth of AI's transition from the "laboratory" to the "industrial sector."

The phrase "lack of full-process transformation" hits the common problem of enterprise AI projects in the past two years — most projects only plug an AI function into a certain step, rather than redesigning the entire process around AI.

Judgment 2: The "China Solution" is not a slogan—it is supported by data

At the Davos Forum, the "China solution" became a high-frequency term, but it was not just empty rhetoric. A few sets of figures:

IndicatorDataMeaning
Proportion of Lighthouse Factories in Chinamore than halfDual advantages of large-scale industrial capacity + innovation ecosystem
Value added of high-tech manufacturing in MayYear-on-year growth exceeded 15%Industrial upgrading is not just empty talk
Monthly production of industrial robotsFirst time exceeding 100,000 unitsPhysical AI mass production window opens
AI plans of 31 provinceshave all been introducedFull coverage of policy infrastructure
Green computing power requirementsThe proportion of green electricity in new computing facilities is ≥80%Not only computing power scale, but also sustainability

I agree with Deloitte China CEO Liu Minghua's observation: China's innovation is transitioning from "point breakthroughs" to "scaled innovation," with an increasing number of cutting-edge technologies leveraging the vast industrial ecosystem and rich application scenarios to rapidly transform into real productivity.

Federico Torti, Head of the World Economic Forum's Centre for Advanced Manufacturing and Supply Chains, put it more specifically:

China's advantage lies in its ability to combine large-scale industrial production capacity with a dense innovation ecosystem. New technologies can be rapidly tested, optimized, and scaled up in real manufacturing environments, supported by an extensive supplier network, engineering talent, and strong industrial capabilities.

In plain terms: China does not lack technology, nor does it lack scenarios; what it lacks is the systematic connection of the "last mile" between the two.

Judgment 3: Three of the top ten emerging technologies are directly related to enterprise digitalization

On June 23, the forum released the "Top 10 Emerging Technologies of 2026" report. Three out of the ten are directly relevant to enterprise CIOs:

Internet of Everything Power Grid

AI-driven energy system optimization scheduling transforms enterprise electricity usage from "passive consumption" to "active participation." For manufacturing CIOs, this is a new lever for saving electricity costs.

World Model

Not just an upgraded version of large models, but a key springboard for AI to go from "understanding text" to "understanding the physical world." For logistics, manufacturing, and warehousing companies, this means AI can predict and optimize physical processes, rather than just handling numbers.

Lattice cryptography

Post-quantum era encryption standards. If your enterprise data assets have a lifecycle exceeding ten years, now is the time to begin assessing the quantum resistance risks of your existing encryption system.

The remaining seven items

Direct lithium extraction, passive radiative cooling, PFAS degradation, precision fermentation, exosome delivery, mRNA vaccines, quantum simulation for drug discovery — these lean toward materials/healthcare/energy, but the demand for digital management at the upstream of the industrial chain is also growing.

An underestimated risk: Hans Hilgenkamp, a professor at the University of Twente in the Netherlands, warned during the forum that the energy consumption issue of AI is severely underestimated, and by 2030, the water usage of AI data centers will be equivalent to the basic living water consumption of 1.3 billion people for a year. If your company is deploying AI infrastructure on a large scale, energy consumption and water resource costs must be included in the total ledger.

Three Practical Suggestions for CIOs

After three days of the forum, my judgment is quite straightforward:

1. Stop only chasing model parameters. Zhu Min made it clear: the bottleneck is not the technology itself, but diffusion and application. Shift the budget from "buying the strongest model" to "bridging the last mile"—data governance, process redesign, and organizational adaptation—these are what determine ROI.

2. Evaluate every AI project against the "full-process transformation" standard. Xu Baoyan hit the mark: AI projects lacking full-process transformation have poor implementation results. If an AI project merely inserts a function into a certain step without redesigning the upstream and downstream processes around it, it is most likely performative AI.

3. Among the top ten emerging technologies, "world models" deserve the most attention. It represents the gateway for AI to move from the digital world to the physical world. If your business involves physical processes such as logistics, warehousing, or production lines, in the next 3-5 years, AI will no longer just help you process data, but will help you predict and optimize operations in the physical world.

关于我们

​我们致力于帮助中小企业实现数字化转型,我们的团队由一群充满激情和创新思维的专业人士组成,他们具备丰富的行业经验和技术专长。

扫一扫获取顾问以及手册

归档
Sign in to leave a comment
OpenAI fixed 32 Linux kernel vulnerabilities, SpaceX sold 80 billion in computing power — enterprise AI infrastructure is being replaced
On June 23, three events coincided. OpenAI's new model identified 32 exploitable vulnerabilities in the Linux kernel within five days; SpaceX disclosed that the total value of signed AI computing power contracts has exceeded $80 billion; IBM announced a $5 billion investment to build an open-source security information sharing platform.