The Data Wire by:
Pure Storage delivers the most advanced data storage platform. We store, manage, and protect the world’s data.
© 2025 Pure Storage, Inc. All rights reserved.
Ron Williams, CEO of Kindo, explains why enterprises must take control of their infrastructure to secure their future with AI.
He describes how AI providers analyze enterprise data for business intelligence and why it poses a threat to intellectual property.
Williams recommends open-weight AI models as a cost-effective and secure alternative to proprietary models.
He suggests adopting a sovereign AI strategy to avoid reliance on external platforms and future resource shortages.
The greatest threat to enterprise value is evolving once again with the growing use of AI. From foreign attackers and malicious insiders to sophisticated new exploits, large organizations have grown accustomed to the rapid fire after years of endurance. Now, some experts say the latest security risk to emerge is one that many leaders still can't see.
Founder and CEO of Kindo, an AI-native terminal for enterprise technical operations, Ron Williams has spent his career on the front lines of hyper growth. Having served as the top security and IT executive at what would become three tech unicorns (Riot Games, Bird Rides, Clover Health), his expertise is backed by over a decade of experience building and defending infrastructure for companies with multi-billion dollar valuations. From his perspective, the C-suite is fundamentally miscalculating the nature of AI risk.
The existential risk of AI: Most AI labs now possess the distribution and capital to launch competing products quickly, precisely, and at scale, Williams says. But in this climate, the conditions are just right for AI partners to transform into predators, often without warning and seemingly overnight. "I have strong feelings on it, and I always say the same thing: not your AI, not your future. I think it's a really important thing to think about and dive into."
Your data, their playbook: Even if AI providers don't train on your data, they do analyze how you use their models to understand your business, Williams explains. With that intellectual property, "They can do pretty much anything they want," from publishing it to using it to build their own competing products. The Anthropic Economic Index is just one of the many smoking guns Williams identifies. "Trust me, these companies are looking at your data very carefully, and they could literally reveal those insights at any time."
The urgency of action: Meanwhile, a collective sense of scarcity is causing many to pick up the pace in the race to AI. Impending resource shortages are quickly closing the window for companies to build their own AI infrastructure before being shut out of the market. Soon, a lack of GPUs and power could force AI providers to ration access, effectively 'picking winners' by deciding who gets compute, Williams continues. "This isn't hypothetical. We're already seeing the impact of rate limits on these models."
Enterprises need a new playbook, according to Williams. What worked for SaaS companies is no longer enough to sustain a competitive advantage. Unlike software, the value of AI-driven applications is in the model, not in the user interface. Without a proprietary workflow to defend, using third-party models is like building on rented land.
The 'sovereign AI' strategy: To protect their business models and create lasting value, enterprises must build and control their own AI infrastructure first, Williams says. "Cede control of your AI stack, and you cede control of your company's future. With AI, the workflow is driven by the technology. There's almost nothing in the user interface. So you're basically just opening up all the doors for providers to see where the early opportunity lies."
Proprietary models are one solution, Williams continues. However, they're prohibitively expensive, often putting companies at a significant cost disadvantage compared to global competitors that are already leveraging cheaper AI.
An open advantage: Most open-weight and open-source models offer a practical, cost-effective, and secure alternative, Williams explains. Often delivering the same value as the best proprietary AI, the advantage of using an open model is that it gives companies more control over their own destiny. "For 90-95% of the current value AI creates for your business, open-weight models can deliver that."
The conversation concludes with an important question: When should a company stop relying on API calls and build its own AI stack? For Williams, the sign of readiness usually takes the form of a simple business metric: value. "Start moving to build your own stack as soon as you identify the use case and see a decent ROI. Once you can start to see the value and after you're convinced that AI is a future part of your stack, you should begin right away."