Inevitable AI Group
SAP paid $1.16 billion for an AI lab that's 18 months old.
Not a decade-old research powerhouse with Nobel winners on staff. Eighteen months. The company is called Prior Labs. It's German. Before this week, most people in the industry hadn't heard the name.
That number is the most honest statement SAP has made about AI in years. They didn't build this internally. They couldn't. Fifty-four years of enterprise software, $35 billion in annual revenue, and the competitive response to the AI moment is a nine-figure acquisition of a company that didn't exist at the start of 2025. You can read it as a bullish bet on AI. I read it as an admission.
This is what happens when your architecture doesn't match the moment. SAP was designed for a world where software stored and processed data. AI doesn't just store and process. It reasons, routes, decides. Retrofitting that into a codebase built on decades of accumulated design decisions isn't an engineering problem. It's a physics problem. You can acquire talent all day and the underlying architecture still won't cooperate.
So they bought the capability. But here's the part nobody is talking about. In the same week SAP announced the acquisition, they also announced they're restricting which AI agents customers can run inside their platform. Approved list only. Nvidia's NemoClaw made the cut. Most things didn't. You want AI in your SAP environment, you use what SAP allows.
Sit with that for a second. They paid over a billion dollars to close the AI capability gap, then simultaneously told customers they can't bring their own intelligence to the table. That's not product confidence. That's a company managing its install base because it can't yet compete on the merits. Buy what you can't build. Lock down what you can't win on product. The playbook hasn't changed since the 1990s. Only the numbers have.
This pattern isn't unique to SAP. It's the standard response when an incumbent realizes its architecture is mismatched to a new era. Acquire capability because building it from scratch inside a legacy system is too slow and too painful. Then protect the install base because the alternative is watching customers walk toward something better. The acquisition buys time. The restriction buys more of it. Neither fixes the underlying problem.
Apple sent a different signal this week. Reports say iOS 27 will let users choose their AI model the way they already choose a default browser. OpenAI, Google, Anthropic — pick from a dropdown. When the most controlled consumer platform in the world starts treating foundation models as interchangeable infrastructure, that debate is over. The model is the plumbing. What you build on top is the product. We wrote last week about how AI agents are replacing the need for connectors entirely — same principle, different layer. The value has moved up the stack.
SAP's $1.16 billion move and Apple's model-selection announcement are telling the same story from opposite ends. One company paying a premium to acquire AI capability it couldn't produce internally. The other making AI models a commodity because the differentiation now lives in the application layer. Both signals point to the same opening for builders who started with the right architecture. That opening has been widening for over a year.
That's the thesis behind Inevitable AI Group. We don't acquire intelligence and bolt it onto existing products. We build companies where intelligence is the foundation from day one. Corebee didn't start as a help desk with AI added later — it was rebuilt from scratch as the AI-native customer support product the market deserves. VScout didn't add an AI layer to a recruiting workflow — it reimagined what recruiting looks like when an agent can own the process from job description to signed offer. Every venture we ship starts from the same question: what does this category look like if you build it for 2026 instead of 2006?
The answer is always different. The starting point is always the same. Clean architecture, AI-first decisions, and a market that's spent years overpaying for software that wasn't built for what it actually needs.
SAP's $1.16 billion tells you the gap is real. The restricted agent list tells you the incumbents know it. They're buying time and buying capability, because the alternative — rebuilding from the ground up — isn't something a $35 billion public company does without significant structural pain. They will keep acquiring. They will keep restricting. And every quarter that passes, the distance between what they shipped and what an AI-native product can deliver grows.
We don't have that problem. We build from scratch every time, by design. That's the model. See the full portfolio at iaig.com.