Artificial intelligence (AI) is more than just a technology—it is poised to become a foundational infrastructure for economies, corporations, and societies alike, said Microsoft CEO Satya Nadella at the 2026 World Economic Forum in Davos on Wednesday.
Speaking with BlackRock CEO Larry Fink, Nadella laid out how AI could reshape productivity, redefine corporate sovereignty, and even serve as a new economic metric for measuring GDP growth.
The discussion centered on one audacious question: “Will GDP growth depend on the number of Tokens?” In Nadella’s framework, Tokens are not just units of code—they are the fundamental building blocks of AI-driven economic output. How efficiently a nation or company produces these Tokens, relative to energy expenditure, could determine its global competitiveness.
Nadella began by framing AI as a historic “Platform Shift”, akin to the emergence of the internet, mobile computing, or cloud technology. Unlike previous innovations, AI offers extensibility and self-transformation, enabling software to reason, adapt, and take autonomous action.
“In the past, programmers wrote code,” Nadella explained. “A document was a document, a website was a website. Now, AI can take a document and turn it into a website, then rewrite that website into an app if needed.” This evolution, Nadella said, has rapidly progressed from tools like GitHub Copilot, which initially only completed code, to fully autonomous Agents capable of managing entire projects 24/7. AI has effectively become an “Infinite Mind” for every knowledge worker.
Drawing on Steve Jobs’ metaphor of the “bicycle for the mind,” Nadella described today’s AI as a hundred-fold amplifier of human cognitive ability, capable of extending expertise and productivity across sectors in ways previously unimaginable.
Addressing skepticism over an “AI bubble,” Nadella drew a clear line between speculative hype and transformative impact. “If AI is just a frenzy for technology companies, then it’s a bubble. If it spreads across industries like electricity and creates real surplus, then it’s a transformation,” he said.
Central to this transformation is diffusion: the equitable and effective deployment of AI across companies, sectors, and regions. Nadella emphasized that without measurable societal impact—improvements in healthcare, education, and public sector efficiency—the technology risks losing the social license to consume energy at scale.
This led to his proposal of a new macroeconomic indicator: Tokens per Dollar per Watt. In Nadella’s vision, the productivity of an economy will increasingly hinge on how efficiently it generates intelligent Tokens relative to energy consumption. “The countries and companies that produce more intelligent tokens at lower energy costs will dominate the future economy,” he said.
Nadella also reframed corporate power in the AI era. Historically, debates around data sovereignty—where data is stored and who controls it—have dominated policy discussions. Nadella argued that the true locus of control lies in model weights: the unique AI parameters a company develops using its own proprietary knowledge.
“Companies that rely solely on external AI models without embedding their tacit expertise into controllable models are leaking their core value,” he warned. In Nadella’s view, corporate sovereignty now means retaining control over AI models that encode the company’s unique knowledge, creating a new type of competitive moat.
The conversation highlighted how AI is transforming organizational structures. At Microsoft, Nadella explained, hierarchical briefing processes for conferences are now largely automated. Using Copilot, cross-functional teams receive 360-degree analyses in minutes rather than hours.
He proposed an “iron triangle” framework for enterprise AI transformation:
Mindset: Leaders must redesign workflows around AI rather than retrofitting old processes.
Skillset: Employees must learn to use, trust, and manage AI, requiring a workforce-wide skills upgrade.
Dataset: Contextual engineering ensures AI receives data embedded with the enterprise’s unique knowledge.
Nadella observed a “barbell effect”: small startups leveraging AI from the ground up can achieve unprecedented efficiency, while large enterprises with rich data can scale massively if they adapt. Medium-sized companies that lag risk losing ground to more agile competitors.
Looking five to ten years ahead, Nadella predicted a “multi-model world” where no single AI system dominates. Competitiveness will hinge not merely on owning a model, but on AI orchestration—the ability to integrate proprietary, open-source, and third-party models with private data to drive business outcomes.
“Orchestration, not ownership, will be the defining capability of leading enterprises,” he said.
A recurring theme was diffusion and inclusion. Nadella emphasized that AI should reach all corners of the globe, not just well-educated or developed regions. Leveraging the existing mobile network and global connectivity, he argued that AI Tokens can now be transmitted more broadly than during the PC or early mobile eras.
He cited a 2023 example: a farmer in rural India used an early GPT-based robot to query agricultural subsidies in his local dialect. “This is the kind of real-world impact that AI diffusion can achieve—extending knowledge and productivity to underserved populations,” Nadella said.
He warned, however, that diffusion alone is insufficient. Skills training is crucial: citizens must acquire AI literacy that enables them to participate fully in the digital economy. Drawing on lessons from the mobile internet, Nadella noted that prior technology waves sometimes created consumption-oriented opportunities without meaningful pathways to employment or economic advancement. AI must do better.
Tokens, Nadella argued, should serve as a new unit for measuring AI-driven economic output. Efficient token production across energy and computing networks will dictate which nations and enterprises thrive. This represents a shift from traditional GDP measures to a model where digital labor and AI output are central to economic competitiveness.
“In a token-powered economy, productivity is no longer linear. It can grow exponentially as AI amplifies human work, reduces wasted time, and transforms services from healthcare to finance,” Nadella said.
The conversation also explored the concept of surplus: AI’s ability to generate output that exceeds current labor and resource constraints. Nadella highlighted that surplus is not about job reduction per se but about freeing human capacity to focus on higher-value work. He stressed the need for responsible diffusion, ensuring that AI benefits are widely shared and do not exacerbate inequality.
“Every society must ask: Are these Tokens improving human well-being? Are we creating equitable productivity gains?” he said.
Nadella concluded with a call to action for both public and private sectors. Effective AI diffusion, combined with infrastructure, talent development, and corporate model sovereignty, will determine global competitiveness over the next decade. The message was clear: the AI era is not just about technology—it’s about economic strategy, societal outcomes, and the very metrics by which we measure growth.
“Diffusion, orchestration, and token efficiency will define the next phase of human productivity,” Nadella said. “Those who understand and implement these principles will shape the economies of the future.”
