At Davos 2026, global leaders spoke about artificial intelligence the way their predecessors spoke about railroads, electricity, automobiles and the internet. AI is now being thought of and treated as one of the largest infrastructure buildouts in history. The combination of power grids, advanced chips and the continual growth of data centers worldwide proves that AI is now a physical, capital-intensive force reshaping and deciding which economies and geopolitical players will win the next decade and beyond.
Despite the waves of excitement and confidence, there were honest concerns about economic imbalance, productivity divergence and who ultimately benefits from this transformation. Those of us gathered in Davos were no longer debating whether AI will change the world and shape our collective futures; that is settled. Instead, our deepest and most profound discussions were around control, governance and timing.
AGI Is Closer Than The Headlines Admit
Ahead of Davos, former OpenAI board member Helen Toner told the U.S. Congress that human-level AI could arrive within one to three years, and that it could pose existential risk. This was barely covered by mainstream news. The majority of the public may still view AI as ChatGPT or something from a science fiction show. Still, the uncomfortable truth that few are publicly acknowledging is that, by almost every traditional measure of intelligence, artificial general intelligence already exists.
AI systems today speak, read and write in more than a hundred languages. They outperform humans on standardized IQ tests, solve advanced mathematics faster, dominate complex strategy games and can query and reproduce enormous amounts of knowledge in seconds. The definition of “general intelligence” keeps shifting because the implications are unsettling. But, we still hire humans, promote humans and trust humans to code and set parameters. This is because it was never intelligence that we as a global community lacked; it was speed and efficiency.
What the world needs, and will continue to need is the ability to understand context, have judgment, accountability and trust. AI only knows that a task exists; humans understand why. People can anticipate unintended effects and recognize when the framing itself is wrong. AI still requires all instructions spoon-fed, prompt by prompt. Humans can see a problem occurring and self-correct before it does. AI corrects only after failure. Humans change their beliefs when reality changes, while AI continues to complete patterns even when those patterns are no longer valid.
Another deeper gap we spoke to at Davos was that yes, AI can be fluent, clever, and even persuasive. But it is not yet a trusted teammate. It does not understand humor, connection or moral intuition.
Superintelligence Itself Is Not The Real Risk
The real AGI risk is not runaway superintelligence, it’s how the world delegates authority and ownership before responsibility exists. Markets are forcing speed and capital investment is accelerating deployment, but institutions are lagging behind in capability. As Elon Musk warned, humans have been the most intelligent beings on Earth for a long time, and that is changing. But intelligence alone has never ruled the world. Power, governance and incentives do.
The central question that must be answered is not when AGI arrives, but which decisions need to be left to the judgment of humans, and why. AGI in some manner is already embedded in markets, codebases, supply chains and governments. Future geopolitical control will not be decided by who has the best AI, but by who sets rigid boundaries before the chance to do so disappears.
Deregulation Leads To Unprecedented Growth
The U.S. economy is now expanding faster than China’s, and the reasons are deregulation, capital and an increasingly aligned relationship between government and technology, which is producing a productivity dividend earlier than most models predicted.
Because of this, geopolitics now lives downstream from AI. Power is reorganizing around compute, energy, chips, data and talent. When top AI engineers command compensation packages once reserved for star athletes and CEOs, it’s a signal that a new economic reality has arrived. This leads to a new way for corporations and global players to align.
Alliances Now Follow Technology Stacks
Another consistent theme in Davos was that alliances are increasingly following technology stacks rather than ideology. U.S. and Middle Eastern technology alignment is emerging while Europe is being squeezed by slower decision-making, heavier regulation and fragmentation. Sanctions are shifting to bottlenecks such as advanced chips, equipment and energy inputs. Countries that successfully convert AI into sustained growth will be the ones that write the rules. Those that don’t will become countries that are dependent upon foreign players.
There were quite a few questions that weren’t addressed on stage but surfaced in many off-stage conversations. Can Europe catch up? What does an AI-defined world mean for Africa? Not all was sunshine and lollipops.
Fracture Beneath The Optimism
Davos revealed a deeper power fracture, as some of the most discussed moments were not optimistic. The tone from parts of the U.S. delegation, including President Donald Trump and Commerce Secretary Howard Lutnick, emphasized leverage over integration and deals over rules. The approach generated attention but also resistance in the form of boos, walkouts and early exits.
From this, an increasing group of leaders are choosing selective cooperation over constant reaction. Not appeasement or escalation, but pragmatic alignment where trust is still possible. The remarks from Canada’s Mark Carney stood out precisely because they avoided illusion and any sort of moral high horse statements. Stability is becoming more and more of a strategic asset.
AI Is Now A CEO Problem
Perhaps the most consequential shift discussed in Davos was not geopolitical, but organizational. AI is no longer an experimental initiative led by IT teams and innovation experts. It has become a CEO-level priority.
According to a BCG presentation during a CEO breakfast at Davos, 72 percent of CEOs now view themselves as the primary decision-makers on AI. Plus, nearly all expect AI agents to deliver positive ROI by 2026. This shows a shift from experimentation to an expectation of revenue generation. Across industries, the topic of change management, not technology, determining who wins was widely discussed. A lot of companies are still trying to retrofit AI into existing workflows, while the genuine visionaries are redesigning entire workflows with AI as the foundation.
Leaders must mandate AI adoption. They require engineers to use AI coding assistants, analysts to start with AI-generated drafts and customer operations teams to be measured on AI-driven resolution and deflection. These mandates and corresponding KPIs that determine wage increases and promotions will be at the base of true organizational transformation.
Infrastructure, Not Models, Is The Bottleneck
AI infrastructure, not models, is becoming a true constraint to fully developed AI implementation and transformation. Training compute has increased almost fourfold in the past year, while inference costs have fallen drastically. The slowdowns are shifting from intelligence to hardwired and manually built infrastructure.
Payment rails for autonomous agents, standardized APIs, developer tooling, energy supply, grid reliability and data center capacity are now the limiting factors. This infrastructure layer will determine how fast AI moves from demos to reliable systems. This is now one of the largest opportunity spaces for startups and governments alike. How can you quickly, safely and affordably solve this problem? Fortunes and empires lie in the answer.
From Globalization to Sovereignty
Commerce Secretary Howard Lutnick’s remarks were very blunt. The current administration believes that globalization, as practiced over the past several decades failed the West. The offshoring of production hollowed out industrial capacity, created strategic dependence on rivals, and left supply chains vulnerable when it came to semiconductors, pharmaceuticals and energy.
The new reality is data and intelligence sovereignty as a strategy. Borders matter and national survival cannot be offshored. Davos has long championed cooperation, but cooperation without resilience quickly becomes dependency. This new era will be defined by regional power blocs, industrial policy as national security and supply chains designed for trust, not just cost.
Quiet Reordering
What ultimately stood out in Davos was not who spoke the loudest, but who demonstrated clarity. The world is being quietly reordered by those who control critical systems, translate AI into actual productivity, build trust and preserve optionality.
Change is arriving faster than most organizations are prepared for and AI still needs to earn the trust of most of the world’s population. Job displacement is accelerating, and there is a growing urgency for governments to respond. Energy and talent now matter as much as chips, and power and compute are strategic constraints.
Global order is not given or simply decided, it is constructed patiently, strategically and often quietly. Davos 2026 made it undeniably clear that AI is no longer a future risk to manage. It is the force reshaping who leads, who follows and who decides what the future of the world will look like and how it will operate.