Davos 2026: The Mood in the Halls Shifts from AI Hype to AI Anxiety

The ideas aren’t the first thing you notice about Davos. The choreography is the cause. Lanyards swinging like metronomes, private drivers sitting outside hotels with their hazard lights blinking in the thin Alpine air, boots knocking slush off the steps of the Congress Center. In previous years, the AI crowd showed up with the smiles of those who thought they were ahead of their time. The smile appeared rehearsed this time, as if it had been maintained for an excessive amount of time.

With artificial intelligence playing the roles of both wonder and suspect, Davos 2026 seems to have evolved from a technology conference to a global accountability gathering. Reuters succinctly portrayed the reality of the hallway: AI and President Donald Trump were the two subjects that kept consuming everything else. The combination was important. Unlike software, AI was not being discussed. It was being discussed like power, the kind that brings product managers and governments together.

It wasn’t that people lost faith in AI that changed. Belief, if anything, became sharper at the edges, but stronger overall. From “possibility” to “rollout,” from idealized prototypes to workflow rewrites and headcount calculations, the terminology changed. Executives discussed speeding up code, compressing projects, and reducing onboarding days. However, the assurance kept running into a more subdued admission: most businesses still don’t know if they have the strength to scale what they’ve tested. Conversations were clouded by this disparity between what the tools could accomplish and what the institutions could take in.

Not on stage were the most illuminating moments. They took place in the side hallways, where the conversation becomes more focused and the lighting is dimmer. The phrase “year of AI ROI” is used in the same way that people say “this will be fine” when they see turbulence hit a plane. According to Reuters, surveys revealed that many CEOs were still not witnessing significant cost reductions or revenue increases, despite leaders’ efforts to promote positive job narratives. The payoff appears to be real, and investors appear to have a convenient faith that it will materialize shortly after the next budget cycle.

Then, with the tone of someone enlarging the frame, Jensen Huang showed up. He made the case in Davos that AI is infrastructure, the kind that necessitates extensive development, not a bubble. The term “infrastructure” took on a different meaning in a town that relies on actual infrastructure, such as heated sidewalks, plowed roads, and tunnels that keep people moving when the snow gets deep. AI ceases to be a private experiment if it becomes infrastructure. It appears to have the potential to lock in winners and losers for ten years, or possibly more, depending on who controls the data centers, chips, and electricity that powers them.

which caused the atmosphere to shift from one of ambition to one of anxiety. Energy kept coming up as a practical limitation—how many models can run, where data centers can be built, and who gets capacity first—rather than as a talking point about climate change. Even though Davos has always enjoyed abstraction, the term “grid bottleneck” is the perfect way to bring a space back to reality. The geopolitical climate, which included trade threats, deteriorating alliances, and the strange feeling that technology leadership now uncomfortably sits close to national security, was also hovering nearby.

However, the most acute tension was related to work, specifically the form of a career rather than the quantity of jobs. As entry-level positions become scarce, young people will be hit by an AI “tsunami,” according to IMF Managing Director Kristalina Georgieva. The tech optimists, on the other hand, maintained that the boom would generate “jobs, jobs, jobs,” particularly in the fields of construction, chips, and energy. The issue is that both may be true. A society can withstand job turnover, but it suffers when the ladder collapses and no one can see the first rung.

It was difficult to ignore how frequently education was brought up as branding rather than redesign while listening to the loop of conversations. Although leaders commended “skills,” the more difficult question remained largely unanswered: what happens if the tasks that were previously used to train judgment are automated first? Lawyers gain knowledge by studying documents. Analysts gain knowledge by working on tedious models. Fixing minor bugs on a large scale helps junior developers learn. The pipeline doesn’t merely narrow if AI consumes that much work; it transforms species.

Hypothetical superintelligence was not the subject of the most sobering Davos moments. They were about public tolerance and harm in the present. Leaders are describing the fine line between adoption and backlash when they speak of “social permission,” not being poetic. The next stage of AI might not be constrained by model performance at all, but rather by whether or not the general public believes the deal is reasonable: increased productivity in return for what exactly—higher costs, fewer entry-level positions, and occasionally unreliable systems in private areas of life?

Politics then permeated everything, louder in private and louder in the plenary. Trump’s dominance as a presence and a variable, with leaders constantly recalculating risk, was not concealed at Davos 2026. As a result, AI began to feel less like an arc in Silicon Valley and more like a strategic tool that, depending on the week, governments will attempt to control, stifle, or weaponize.

Davos’ message wasn’t clear-cut. A change in facial expressions was conveyed. It appears that the era of simple AI storytelling—the glitzy demonstrations, the talk of “inevitable progress”—is giving way to something more mature and uncomfortable: discussions about legitimacy, labor ladders, governance, and energy constraints. In those corridors, the future didn’t seem so far off. It felt real, like it was already happening, like someone was being charged for the expenses.