The alpha, twice a week
That was 3 min.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
Inside Blackstone · 3 min · Nov 3, 2026
Token spend needs its own controls
Eric Glyman
Eric Glyman says falling AI costs are driving more use, but companies are learning to match models to tasks.
The episode’s best minutes
Full analysis- 10:30AI-intensive companies are growing and hiring faster5 min
- 15:42Token spend needs its own controls3 min
- 3 more moments in the full episode
“Many CFOs are lacking kind of the core capabilities to manage this third pillar of spend.”
Eric Glymanat 18:33 · Inside Blackstone
Takeaways
- Use the right model for each task instead of overpaying for capability.
- Agentic AI costs can vary from cents to dollars per request.
- CFOs need new tools to manage AI spending.
About this segment
Eric Glyman says falling AI costs are driving more use, but companies are learning to match models to tasks. He describes tokens as a third spending pillar that CFOs need new ways to manage.
Eric Glyman is cofounder and co-CEO of Ramp.
Co-founded Ramp, now used by 85,000 businesses and building AI tools to automate finance and manage company spend.
- CA
Christine Anderson, Host · Global Head of Corporate Affairs, Blackstone
From “Ramp Co-Founder and Co-CEO Eric Glyman on AI Adoption, Jobs, and Token Spend”, Nov 3, 2026. Watch the full episode
Thanks. We'll look at and fix what's off.
Transcript
From the transcript Inside Blackstone publishes with the episode.
- 15:47
This is essentially what companies are spending for their AI usage. That's right. I think it's something that we're gonna be talking a lot more about in the future. What is your data pointing to and where do you think this goes? The last two months have been among the most fascinating in all of the data. So, for many years, what you would hear in the narrative about this was this idea of Jevons paradox, he’s an economist from, I think, 100 plus years, who observed that if you reduce the price of certain commodities that the increase in consumption, the increase in quantity demanded, could more than offset.
- 16:24
Put differently, you discount, you sell a lot more. And that was happening for multiple years and you'd see the explosion in revenues of these companies. These models are incredible for coding, but many companies today who've used it saying I'm paying for more stuff and maybe they've made individuals somewhat more productive but the companies still have bottlenecks and so, CFOs are saying hey we're in our ROI era and we need to see the return on this before that. You’re starting to see companies saying, okay, wait a minute, I want to use AI, but maybe I don't need an Olympic athlete to walk my dog.
- 17:00
I don't need to rent a Ferrari to pick up groceries. Right. And so, you're staring to see businesses assess, how do I use the right model for the right job? And well, I think it's an incredible opportunity for businesses. It's led to both excitement, but also new questions for the labs that weren't there before. So you talk about token spend as being the third pillar of business spending. Yes, yeah.
- 17:21
So obviously you're saying that companies are gonna start to really have to evaluate how they spend… Yeah. … money on tokens and where and to what task? That's right. And if you think about it from a CFO perspective, you start with there's a company and an idea and there's some money and then, what happens? You can hire people, classically, or you can hire a vendor, go take care of stuff. And that was kind of about all you could pick. The first two pillars?
- 17:46
That's it. The first two pillars. And intelligence came on. And in some sense, it's not quite like the first or the second. In some sense, if you look at most of the dollars are going to spend, that's called agentic. You ask a question, and there's a session that gets spun up. And a computer basically hires an agent, it may hire another agent, they go debate it out. And a question could cost you 10 cents, $ 1, $ 20.
- 18:12
Who knows? It depends. It's almost like you hire a contractor to fix something in your house. You don't know what the bill will be. So in some sense, it's kind of like you're hiring people, but it’s, in other ways it's like a vendor relationship. You don't own the intelligence, you're renting it. This is an outside party. And so, while this is going on and businesses want to spend more on this and we see in our own data spend on AI grew 21 times over the last year.
- 18:38
Many CFOs are lacking kind of the core capabilities to manage this third pillar of spend. And so, I think it's new and different. It will be an input for how companies operate, but it's not well understood about the best way to manage this kind of spend.