What happens when you turn AI computing into a financial asset?

AI Video & Visuals


00:00 Speaker A

So, in its simplest form, computing is processing power. And it’s basically processing power, especially for AI-type services. So

00:08 Speaker A

If you’re a big cloud company, or a neo-cloud like Core Weve, you’re basically selling computing. And if you’re a large user, maybe an AI lab, or any other company that needs AI services right now, you’re a compute buyer.

00:22 Speaker A

And it’s clear that there could be a kind of financial market, a kind of futures market, around this.

00:28 Speaker A

It’s very difficult to get it right. This is not a homogeneous asset.

00:34 Speaker B

right.

00:34 Speaker A

It’s really, really difficult to turn this into a financial asset, but there’s a lot of money being spent on this now and the appetite to make it happen is clear.

00:41 Speaker A

Many people are trying. I think it will work out somehow in the end. The question is: Will the current wave work, or will we have to wait much longer for a kind of financial market to emerge that matches the actual physical market underlying it?

00:54 Speaker B

In other words, to your point, the products for which there is a market are homogenized.

01:00 Speaker B

We think about oil, we think about agriculture, we think about metals, right?

01:03 Speaker B

The grade may be different, but that’s okay.

01:05 Speaker A

you do.

01:05 Speaker B

But, but they, at the end of the day, oil is oil.

01:09 Speaker A

This is the discussion I have with people all the time. When I talk to startups and companies that are trying to financialize computing, they always say, oh, those assets aren’t homogeneous, right?

01:17 Speaker A

Singapore buys a barrel of jet fuel, and the Midwest buys a barrel of crude oil.

01:23 Speaker B

However, there are different markets for crude oil jet fuel.

01:24 Speaker A

Yes, it certainly does. But they say even though you know there are a lot of commodities and they’re flying around, you can still hold Brent crude oil, you can still hold WTI, you can still hold benchmarks.

01:34 Speaker B

Understood.

01:34 Speaker A

Now, I think the reality is that when it comes to computing, we’re talking about something much less homogeneous than something like that.

01:41 Speaker A

What you’re talking about is that whether you’re renting a chip like Nvidia’s Hopper chip or an H100 chip, or you’re renting the same chip from different providers, prices can vary by a factor of 2, 3, 4, not just in different parts of the world, but even within the same region of the world.

01:58 Speaker A

right?

01:58 Speaker B

What determines its price?

02:00 Speaker A

Many things. All sorts of fundamentals like the stack, the actual physical housing of the chip, the reliability of the data center.

02:12 Speaker B

What kind of power is it connected to?

02:13 Speaker A

absolutely. Yeah. All of this is very important. Well, that’s something financial market investors know nothing about, right?

02:20 Speaker A

Therefore, we need clean and simple benchmarks. There are companies trying to do this on silicon data using proprietary indexes.

02:27 Speaker A

The reality is that it is very, very difficult.



Source link