What happened
Microsoft CEO Satya Nadella published an essay warning that a few frontier AI models risk absorbing industry expertise and commoditising it, potentially "hollowing out entire industries." Nadella introduced "human capital" and "token capital" concepts, advocating for businesses to build and own their AI capability, decoupling institutional intelligence from specific models. This coincides with a class-action lawsuit alleging Microsoft concealed slowing Azure growth and escalating AI infrastructure costs, with Q2 capital spending reaching $37.5 billion, a 66% year-on-year increase. Microsoft also cancelled most internal Claude Code licenses for its Experiences and Devices division.
Why it matters
AI centralisation demands sovereign data strategies from procurement teams and platform engineers, who must prioritise developing proprietary learning loops to maintain institutional knowledge and avoid vendor lock-in. Microsoft's Q2 capital spending of $37.5 billion, a 66% increase, and the cancellation of internal Claude Code licenses underscore significant financial and strategic pressures. This follows research warning of global inequality risks from concentrated AI development, reinforcing the need for decentralised, sovereign AI strategies. Teams should assume high AI infrastructure costs will persist and invest in mechanisms that decouple their unique expertise from transient model dependencies.
