Excited to go on @MTSlive at 1PM pt! nitter.cf/MTSlive/status/2098080…

3 root causes of "The AI Fragmentation Tax" 1) Visibility is fragmented: About half of companies don't track LLM API costs at all, even when AI is core to the product. 2) Forecasting is broken: Only 15% of companies can predict AI spend within ±10% accuracy, which makes budgeting and margin protection nearly impossible. 3) Infrastructure is sprawling: Agents, MCP servers, LLMs, APIs, and event streams all connect across on-prem, cloud, and hybrid environments. Every touchpoint generates cost that's invisible in isolation but huge in aggregate. Even companies not integrating AI into their product directly are still heavy consumers of third-party LLMs, so they add cost with no measured revenue to offset it. The fragmentation tax compounds the longer it exists. Every release cycle makes it worse if you don't address the 3 root causes. That's why we're obsessed with building a unified cost governance layer for our customers. As AI speeds up the convergence of many capabilities, margin discipline becomes a differentiator.

I meet with a lot more CFOs than I used to. We didn't always see C-suite so involved in a technical platform like Kong, but now it's more common because everyone needs to know where the money is going with their huge AI bills. An AI Gateway becomes one of the best tools to bring financial clarity to the tech investments your org is making, but the conversations can be difficult for people from different worlds of finance and tech. We made this "Operational Metrics Translation Table" to help technical and finance teams have more productive conversations. Your CFO doesn't need a lesson in p99 latency or token throughput, they need to know what that metric means for the P&L. Every operational metric your gateway produces has a financial equivalent and the table maps them directly. 4 questions your gateway can help you answer: What's our gross margin per AI interaction, by customer segment? What's our cost exposure if a major LLM provider reprices tomorrow? How much margin is our gateway already creating through caching and routing, and is anyone measuring it? Which internal teams are driving our AI spend, and do they know it?

Those who control the mountain pass win the war. Sun Tzu