Good morning. Earlier this week, I got a message from a colleague in IT, who gave me a heads up that they’d loaded more credits into Perplexity so I could “pull them for my needs.” And pull them, I did, immediately creating an internal report and a research deck for an upcoming meeting. Score! Much like securing an H1B visa, back when it didn’t cost $100,000 a pop, getting access to AI tokens or credits these days can feel a little like winning the lottery.
Welcome to the age of tokenomics. Nvidia CEO Jensen Huang is now giving his engineers tokens equivalent to half their salary, saying that he’ll “go ape” if they don’t use at least that much. Stripe CEO Patrick Collison has warned of “token pilfering.” Some CEOs are using tokens to track how their teams are deploying AI, from the productive tokenmaxers overhauling their departments to the tone-deaf dabblers squandering compute on useless projects. (Turns out you don’t need a 15-page report to pick a lunch spot in Dallas.) Once people are hooked, it’s time to make them pay. I recently spoke to two tech CEOs on the front lines:
Teradata CEO Steve McMillan: “Some of our lead developers will use their entire monthly token allocation in like three hours … My teams are thinking of it very much in terms of people equivalency: ‘Where do I deploy my capital to maximize return between agentic spend, tokens, and human spend?’ I think we almost need a driving license before you’re let loose in terms of being able to utilize all of this technology in an unharnessed way. I don’t let my kids loose with my credit card … If you’ve got a Formula One driver, you’ll trust that individual to get the best out of the tooling. It’s going to be expensive, but it’s probably going to be worth it, compared with somebody on their normal commute. I use three words when I think about the AI environment. One is trust, can you trust that solution? Next, is it ethical? Do you understand how that solution got to that answer? And is it efficient?”
