Steve Hou, Head of Research at Silicon Data, joins Ben Lorica to unpack the data behind the AI compute market: GPU rental indices, forward curves, and token expenditure trends that reveal a market still tightening despite talk of oversupply.<br>Subscribe to the Gradient Flow Newsletter 📩 https://gradientflow.substack.com/<br>Subscribe: Apple · Spotify · Overcast · Pocket Casts · AntennaPod · Podcast Addict · Amazon · RSS.<br><br>Detailed show notes and transcript, can be found on The Data Exchange web site.

The Bloomberg Terminal for AI Compute

The Data Exchange with Ben Lorica

All right, so today we're in for a treat. We are with Steve Ho, Head of Research at Silicon Data. Website is silicondata.com. Tagline is Compute Market Intelligence for Better Decisions, Real-Time Prices, Transparency and GPU Performance, Data for Traders, Financial Institutions and Builders. So Steve, welcome to the podcast. Thank you very much, Ben, for having me.

So we will have a kind of interesting discussion about the current state of AI. But before we do that, let me attempt, Steve, to describe to our audience and feel free to correct me at any point the types of data you guys collect. So this will put in context some of your opinionating during the course of this podcast. So one data set is an independent benchmark, or I guess these are all

independent, but GPU rental price index, right? So for what's the cost of renting a GPU, you have a RAM index as well. You have LLM token expenditure index, which is a blended rate in dollars per million tokens for what it actually costs to run inference. And then one of my favorite things is this GPU forward

curve, which is like a yield curve for GPUs. But that might be too complex for our audience there. But anyway, I love talking about simplifying complex things. So happy to talk about all of them. But just generally, I'm trying to paint the picture here that you folks collect data for the cost of hardware, and then the cost of running inference. Is that a fair description of what you do? Or am I

The Bloomberg Terminal for AI Compute

Read the full transcript

The candid version is on a podcast,
not the earnings call.

The Bloomberg Terminal for AI Compute

Read the full transcript

The candid version is on a podcast, not the earnings call.

The candid version is on a podcast,
not the earnings call.