Six months ago, we’d walk into CTO conversations and hit a wall. “Latency,” they’d say. “Your gateway adds latency. We can’t accept that.” We’d show benchmarks, explain optimizations, talk about marginal milliseconds. Didn’t matter. Latency was the objection that killed deals.
We’re not seeing that anymore.
Over the last month, we’ve had a string of gateway conversations with CTOs at mid-market and enterprise companies that tells a different story. The latency conversation isn’t gone, but it’s not the dealbreaker it used to be. Something else has moved to the top of the priority stack: cost visibility, token control, and how an AI gateway fits into their existing Microsoft and MCP infrastructure.
Take a recent call with a CTO at a 2W company running production AI workloads. He brought up latency first, which we expected. But here’s what made this different: he didn’t use it as a blocker. Instead, he said they’d previously written off gateways entirely because of latency concerns, but now he wanted to see if that had changed. When we explained our current performance metrics, he agreed to a 15-minute follow-up in a Microsoft stack environment to test actual behavior. The call ran 687 seconds. He was listening.
What changed his mind wasn’t our latency story. It was budget anxiety. His team is running multiple AI models across different providers, and they have no visibility into token spend. No way to set controls. No guardrails. He’s building custom MCPs to work around the problem, which is expensive and messy. An AI gateway that gives him cost transparency and token limits suddenly became relevant to his actual operational headache.
We saw the same pattern with a CTO at an enterprise level, an advisor to several Blackstone portfolio companies. He spent 575 seconds on a call recognizing the exact value of a gateway for cost visibility and token control. He’s now pulling in multiple portfolio CTOs because the problem isn’t unique to his company. It’s systemic. Every CTO running AI in production right now is struggling with the same thing: cost is opaque, and they need to control it.
The numbers matter here. These conversations are longer than we’d expect for an initial “not interested” call. We’re seeing prospects willing to invest time when the conversation shifts from latency to operational cost. They’re inviting peers. They’re scheduling follow-ups.
This is a fundamental shift in how technical buyers evaluate gateway solutions in 2026. Latency still matters, but it’s table stakes now, not a dealbreaker. The real buying criteria have moved to cost opacity, token governance, and integration with existing stacks (Microsoft, MCPs, custom tooling). CTOs are building workarounds because their current vendor stack doesn’t give them what they need. A gateway that solves that problem is worth a conversation.
The lesson for us: stop leading with latency benchmarks. Start with cost and control. Ask CTOs about their token spend, their MCP investments, their visibility into AI costs. That’s where the pain is. That’s where the deal is.