Microsoft CEO Satya Nadella is warning organizations to look beyond the cost of using AI and consider what they may be giving up in return. As enterprises expand AI adoption, he argues the conversation should include not only what AI delivers, but also the value of the knowledge organizations contribute through everyday use.
His comments add another consideration for organizations investing in AI. Enterprise AI discussions are beginning to include the knowledge employees contribute through everyday use, along with questions about who ultimately benefits from it.
This discussion comes as many organizations move from AI experimentation to long-term deployment. Decisions made today could shape ownership and control of enterprise knowledge for years to come.
Why It Matters: The discussion around enterprise AI is becoming less about access to powerful models and more about the knowledge that helps those models improve. As organizations rely more on AI in day-to-day operations, understanding where that knowledge goes and who benefits from it could become just as important as selecting the technology itself.
- The Cost of AI May Go Beyond the Invoice: According to Nadella, organizations effectively “pay twice” for AI. The first cost comes through subscriptions, API calls, or token usage. The second comes from the business context employees provide while using the technology. Beyond subscription and usage costs, they also contribute the business context that helps AI perform more effectively.
- Everyday AI Use Can Reveal Institutional Knowledge: AI models improve through what Nadella calls customer “exhaust,” including prompts, tool usage, and especially user corrections. Those interactions capture institutional expertise that has been built through years of experience. In his view, that knowledge represents intellectual property a competitor could never simply acquire.
- Data Ownership Is Becoming a Bigger Part of AI Strategy: A major concern is AI providers that reserve the right to learn from customer interactions. Nadella advocates keeping ownership of prompts and feedback within proprietary learning environments while using orchestration layers or AI gateways that allow organizations to move between models without becoming dependent on a single provider.
- The Debate Over AI Training Is Expanding: Another issue is whether AI providers should be able to train on publicly available information while preventing customers from using techniques such as model distillation to learn from commercial models. Nadella views that imbalance as inconsistent, arguing organizations should have greater opportunity to benefit from the intelligence they help create through AI use.
- Enterprise AI Is Moving Toward Greater Control: Although he never explicitly recommends open-source models, Nadella’s recommendations closely match decisions already being made across the enterprise. Solo.io reports customers moving from proprietary models to open-source alternatives running on-premises, while Vercel and OpenRouter have seen growing demand for open-source models. Together, those developments suggest many organizations are placing greater value on controlling how enterprise knowledge is retained and used.
Go Deeper -> Satya Nadella has issued a shocking warning to companies using AI – TechCrunch
Satya Nadella has issued a shocking warning to companies using AI – MSN
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
