AI Is Scaling Faster Than the Systems That Support It 

One message has been hard to ignore at Davos this week. AI has moved into business operations faster than the infrastructure and guidelines needed to support it at scale.  

That gap is especially visible in supply chains. AI can support fast decisions in moments of disruption, and our Global Trade Observatory Annual Outlook Report shows that 43% of logistics leaders see it as a key growth driver. But the system can still break down at the seams. A striking 60% of leaders point to customs and border frictions as major sources of delay, showing how disconnected data keeps insight siloed and limits forecasting.

Unfortunately, this is more than a tooling problem; it’s an integration issue. As the World Economic Forum’s MINDS (Meaningful, Intelligent, Novel, Deployable Solutions) programme shows, organisations that scale AI successfully embed it into core systems and operating models, rather than layering tools onto fragmented foundations. When that integration is missing, progress can look impressive at first, but it rarely holds up under pressure.

And governance can’t be an afterthought. As Kate Kallot warned during the Centre for Frontier Technologies and Innovation ‘‘The Intelligent Co-Worker ‘’ panel, if models aren’t grounded in real-world contexts, “they will make confident decisions based on very incomplete records.” That same message has come up repeatedly: once AI systems are live and making decisions, adding guardrails later is not just difficult; it’s expensive, and trust is hard to recover.

Attention is also turning to durability. Even though quantum computers aren’t here yet, their huge promise is already shaping how organisations think about the future. With uncertain timelines but clear implications, leaders are rethinking encryption and long-term data security. As Kimberly Budil, Laboratory Director, Lawrence Livermore National Laboratory warned, attackers could be “gathering data up now and waiting for the moment when these encryption methods are able to be broken to look at all that old data.”

In global trade and wherever else this technology is applied, it can only create real value when systems work together, and when trust and responsibility are built in from the start. With so many public and private platforms operating under different standards, innovation only sticks when it’s treated as shared infrastructure, not a bolt-on, and designed to last. In the end, innovation will be judged less by how quickly it’s adopted, and more by how well it holds once embedded in the everyday workings of global trade.