The Difference Between Using AI and Building AI
10 min read
Jan 15, 2026
Naeem's talking head
“So it cost you $700 because of the tokens?”
This question from a technology management consultant stopped me cold. I had just explained that producing a 6-minute demo video of my AI-powered application cost $700. His assumption—that I was paying per-token to some AI provider—revealed a fundamental misunderstanding about what I had built.
“No,” I explained. “I don't pay for tokens. I built my own AI. My cost is compute—$50 per hour for the GPU servers that run my models.”
His response: “Yeah, we have an API key that we use to hit a server and interact with AI.”
Maybe he didn't understand the significance of what I had built. Even after I explicitly stated that I had built my own AI, he seemed only knowledgable of consuming someone else's. This wasn't a failure of explanation—it was a gap in understanding that exists across much of the technology industry. If a technology consultant doesn't grasp the distinction, how many others share this blind spot?