Axios C-Suite: How to control your company's brain
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Brendan Lynch/Axios
AI is becoming the operating system for how companies think, sell and decide. We need to view this — and protect this — like the proprietary brains of our businesses.
Why it matters: Palantir's Alex Karp warned on CNBC that big AI companies could use your company's data to learn from you and, one day, squeeze you.
- Karp said every company using AI should be able to answer a few basic questions: "Who owns the data? Where is it cached? Are the prompts secure? Is this being transferred to you? Are you being copied?"
Just ask Figma: The Information reported last month that Anthropic told the design firm its Claude Design product would be a complementary tool, then added competing features days before launch.
- Figma's CEO said Anthropic wasn't "consistently candid" during development.
Get familiar with "alpha" (the knowledge that separates you from the market) and "sovereign AI" (controlling the models you use and proprietary apps connecting them to your alpha). These will dictate how you use AI going forward.
Here's how to think it through:
- Map your alpha. Sort your AI work into two piles: commodity tasks (drafting, coding, research, things anyone's model can do) and crown jewels, or your alpha. Most CEOs have never made this list.
- Own the application layer. Rent frontier models for commodity work. Run open-weight models you control for crown-jewel work. Sovereign AI means controlling the layer where the model meets your alpha, or you may be trusting a vendor not to become your competitor.
- Control the switchboard. OpenAI and Anthropic both say they won't train on business customer data by default. But the real power over your data sits in the router — the software layer that acts as the switchboard for which model handles which task, based on cost and security. Palantir and Databricks are both racing to sell you that layer. Whoever controls your router controls your alpha exposure.
- Watch the salesmen. Karp went on CNBC to promote Palantir's Nvidia deal — and ended up hawking his product. Anthropic CEO Dario Amodei called the emergence of free open-weight models a "serious concern" to safety, but they're a serious concern for his business model, too. Everyone's selling something. Ask the same questions of all of them.
The bottom line: We need to understand sovereign AI because it captures all the big, messy fights ahead: trust in big AI, open versus closed models, data protection, costs and creating moats labs can't replicate.
