top of page

Satya Nadella at Davos and the geopolitics behind AI “token factories”

When “tokens per dollar per watt” becomes power, AI stops being a tech story and starts becoming energy, trade, and national strategy.





Satya Nadella, the CEO of Microsoft, at his recent Davos conversation, was framed as an AI story, but the subtext was geopolitical. He described “token factories” as the new industrial base: data centres, chips, networks, and the electricity that turns raw compute into usable AI output. That single metaphor forces a sharper question: who controls the inputs that make modern intelligence cheap and abundant, and who becomes dependent on someone else’s grid? 


In a world where AI is increasingly baked into productivity, education, and public services, access is not just commercial. It becomes a national capability. And for South Africa, where energy constraints and uneven digital infrastructure remain real, Nadella’s argument lands uncomfortably close to home.


CONTEXT AND BACKGROUND

The last two decades taught us that digital advantage compounds. Countries with reliable connectivity, affordable devices, and scalable cloud infrastructure built faster-growing tech sectors and more resilient economies. AI now accelerates that same dynamic, but with a twist: the bottlenecks are physical again. Electricity supply, grid stability, data centre permitting, chip supply, cooling, and capital expenditure determine how much AI capacity a country can actually run.


Davos this year made that convergence hard to ignore. AI was everywhere, but so was politics: trade rules, sovereignty debates, and the strategic competition behind “who gets to build what, where”. Reuters captured that mood clearly, with AI and politics dominating the week’s conversations.


INSIGHT AND ANALYSIS

Nadella’s most useful contribution is that he pulled AI back down to earth. “Tokens per dollar per watt” is not a slogan for tech Twitter. It is a competitiveness metric that ties GDP aspirations to energy pricing, infrastructure delivery, and industrial policy. If you cannot produce tokens cheaply, you will buy them. And if you buy them, you inherit someone else’s rules, pricing power, and strategic leverage.


This is where the geopolitics become unavoidable. Token factories require chips and specialised hardware at scale. They require steady power. They require water and land. They require permitting and community acceptance. That means domestic policy choices and international relationships suddenly determine whether your local firms get “infinite minds” on tap, or whether AI remains an imported service that sits above your economy rather than inside it.


There is also a warning embedded in Nadella’s message: even if tokens are technically available everywhere, the economic benefits will not be. The Financial Times reported Nadella’s concern that the AI boom could look like a bubble if adoption and gains stay concentrated in big tech and wealthy countries. That is not just a moral point. It is a stability point. Uneven diffusion breeds political backlash, protectionism, and distrust.


Finally, there is the “social permission” problem. TechRadar highlighted Nadella’s argument that society will not tolerate scarce resources being diverted into AI indefinitely unless the outcomes are real: better health, better education, better public services, better productivity. That changes how leaders must talk about AI: less spectacle, more measurable benefit.


IMPLICATIONS

For South African policymakers, this reframes AI strategy. It cannot be limited to regulation, ethics, or skills in isolation. It must include energy resilience, data centre policy, spectrum and connectivity, and investment pathways that do not lock the country into permanent dependency. If tokens become the new utility, then the politics of utilities apply: access, affordability, reliability, and national interest.


For business leaders, the lesson is to stop treating AI as a procurement decision and start treating it as operational risk and competitive positioning. If your AI capacity is priced in foreign currency, tied to external infrastructure constraints, or subject to abrupt policy shifts elsewhere, your transformation roadmap is fragile. Fortune noted Nadella’s warning that firms must reinvent knowledge work and workflows for AI to create real value, not just spend.


For educators and parents, the issue is even more direct. If token access and costs shape productivity, then they will shape opportunity. Schools that can afford AI tools, connectivity, and teacher support will widen the gap from those that cannot. AI literacy becomes a civic issue, not just a career advantage.


CLOSING TAKEAWAY

Nadella’s “token factories” metaphor is a reminder that the future will be built as much by planners, engineers, and energy strategists as by software developers. The next phase of AI will reward countries that can turn scarce power and capital into broadly shared capability, and it will punish those that outsource the foundations. South Africa should treat this moment as a practical call to action: stabilise energy, modernise digital infrastructure, build skills that translate into real jobs, and insist that AI deployments earn trust through outcomes. The geopolitics are already here. The question is whether we arrive prepared or arrive dependent.


Author Bio: Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net

 
 
 

Leveraging AI in Human Resources ​for Organisational Success
CTU Training Solutions webinar

bottom of page