WEBVTT - Michael Intrator Talks Data-Center Delay, Supply Chain

0:00:02.520 --> 0:00:08.600
<v Speaker 1>Bloomberg Audio Studios, podcasts, radio news. Let's talk about where

0:00:08.600 --> 0:00:10.920
<v Speaker 1>we are in this cycle, because core Weave shares, as

0:00:11.080 --> 0:00:14.000
<v Speaker 1>Ed pointed out earlier, under pressure. That's after it lowered

0:00:14.000 --> 0:00:16.320
<v Speaker 1>its annual revenue forecast due to delay at the third

0:00:16.360 --> 0:00:19.880
<v Speaker 1>party developing a data center really overshadowed impressive growth in

0:00:19.920 --> 0:00:22.640
<v Speaker 1>revenue in order backlog. Let's get straight to it with care.

0:00:22.680 --> 0:00:26.360
<v Speaker 1>We've CEO Michael and Trader. How frustrating is it that

0:00:26.400 --> 0:00:28.880
<v Speaker 1>you don't own your entire supply chain? How much of

0:00:28.920 --> 0:00:31.680
<v Speaker 1>an issue is the sort of power you have to

0:00:31.680 --> 0:00:32.640
<v Speaker 1>give to third parties?

0:00:33.360 --> 0:00:37.040
<v Speaker 2>Oh so, look, look, I think it's important to understand

0:00:37.040 --> 0:00:43.080
<v Speaker 2>that every single part of this ecosystem is dependent upon

0:00:43.159 --> 0:00:48.160
<v Speaker 2>other parts of the ecosystem. Nobody controls the entire supply chain,

0:00:48.479 --> 0:00:52.760
<v Speaker 2>and that's just the nature of doing business at this

0:00:52.880 --> 0:00:57.360
<v Speaker 2>scale on projects that are as complicated this, whether it's

0:00:57.480 --> 0:01:03.720
<v Speaker 2>you know, you know, the design and architecture of the chips,

0:01:03.920 --> 0:01:06.400
<v Speaker 2>through the fabrication of the chips, through the building of

0:01:06.440 --> 0:01:08.480
<v Speaker 2>the data center, through the concrete that has to go

0:01:08.520 --> 0:01:10.800
<v Speaker 2>into the data center, like it's just, you know, like

0:01:11.080 --> 0:01:15.160
<v Speaker 2>it's really hard to even contemplate a world in which

0:01:15.400 --> 0:01:18.280
<v Speaker 2>you have true control over every aspect of it. And

0:01:18.360 --> 0:01:21.680
<v Speaker 2>so you're going to work with partners, and you're going

0:01:21.720 --> 0:01:23.800
<v Speaker 2>to work with good partners that are able to help

0:01:23.840 --> 0:01:26.840
<v Speaker 2>you drive your business forward.

0:01:27.480 --> 0:01:28.440
<v Speaker 3>Is this a good partner?

0:01:28.600 --> 0:01:30.479
<v Speaker 1>She's had issues at this PowerShell level.

0:01:30.600 --> 0:01:32.720
<v Speaker 2>Yeah, Look, they are a good partner, right Like we've

0:01:32.720 --> 0:01:36.000
<v Speaker 2>been working with these guys since twenty and eighteen. There

0:01:36.720 --> 0:01:39.800
<v Speaker 2>has been a stumble, there's an issue. The issue is

0:01:39.840 --> 0:01:43.920
<v Speaker 2>going to be cleaned up rapidly. But I wouldn't say

0:01:43.959 --> 0:01:45.760
<v Speaker 2>that they're not a good partner. I think that, you know,

0:01:46.080 --> 0:01:48.680
<v Speaker 2>we are. You know, we've gone back and contracted with

0:01:48.720 --> 0:01:51.520
<v Speaker 2>them again and again, and you know they've contracted with

0:01:51.600 --> 0:01:53.880
<v Speaker 2>us again and again because you know, we do work

0:01:53.880 --> 0:01:56.720
<v Speaker 2>well together. It's just you know, occasionally you hit a

0:01:57.160 --> 0:01:59.920
<v Speaker 2>bump in the road. You know, what we're talking about

0:02:00.040 --> 0:02:04.760
<v Speaker 2>year is a contract that gets pushed back by you know,

0:02:05.800 --> 0:02:08.880
<v Speaker 2>one quarter, let's say, and that the impact of that

0:02:09.280 --> 0:02:12.760
<v Speaker 2>on our revenue is a delay of revenue, not a

0:02:12.760 --> 0:02:15.440
<v Speaker 2>loss of revenue. And uh, you know, I think that

0:02:15.600 --> 0:02:18.440
<v Speaker 2>is the definition of working with an ecosystem of good partners.

0:02:19.600 --> 0:02:22.760
<v Speaker 4>Michael, good Morning. Which customer was it pleased that was

0:02:22.800 --> 0:02:24.120
<v Speaker 4>impacted by this delay.

0:02:24.400 --> 0:02:28.160
<v Speaker 2>We don't speak to specific customers within the data centers.

0:02:28.200 --> 0:02:30.200
<v Speaker 2>That's that's not how we look at it. But you know,

0:02:30.240 --> 0:02:34.440
<v Speaker 2>we we we really are working with the whole you know,

0:02:34.840 --> 0:02:38.960
<v Speaker 2>or a material part of the broader.

0:02:38.720 --> 0:02:39.200
<v Speaker 3>Kind of.

0:02:41.480 --> 0:02:44.800
<v Speaker 2>Ecosystem that consumes compute at this scale. So you know

0:02:44.880 --> 0:02:47.600
<v Speaker 2>that that's you know, there's a large project, it's from

0:02:47.639 --> 0:02:51.640
<v Speaker 2>a large counterpart, and you know, like I said, you

0:02:51.680 --> 0:02:54.760
<v Speaker 2>know they they understand uh the impact of this delay

0:02:54.800 --> 0:02:59.280
<v Speaker 2>and that they are shifting with us the contract back

0:02:59.320 --> 0:03:02.800
<v Speaker 2>to ensure that total contract value that was being contemplated

0:03:02.919 --> 0:03:04.400
<v Speaker 2>is going to be captured at this railm.

0:03:06.360 --> 0:03:08.440
<v Speaker 4>I thought you might say that, but I wanted to

0:03:08.480 --> 0:03:11.440
<v Speaker 4>offer you the opportunity to explain who the customer was. Anyway,

0:03:11.760 --> 0:03:14.480
<v Speaker 4>I appreciate this is going to sound a little bit

0:03:14.560 --> 0:03:17.000
<v Speaker 4>like I work in your internal audit team or in

0:03:17.040 --> 0:03:19.799
<v Speaker 4>your risk team, But there is a bigger issue here

0:03:19.800 --> 0:03:22.800
<v Speaker 4>of managing third party risk as much as you can.

0:03:23.400 --> 0:03:27.720
<v Speaker 4>So if your supply is tight today, what procedures are

0:03:27.760 --> 0:03:30.799
<v Speaker 4>you putting in place for the event that your backlog

0:03:30.880 --> 0:03:33.600
<v Speaker 4>doubles and supply is even tighter down the road.

0:03:34.440 --> 0:03:36.960
<v Speaker 2>Listen, that's a that's a that's that is the question

0:03:37.080 --> 0:03:40.720
<v Speaker 2>that we work on across our company continuously, right and

0:03:41.120 --> 0:03:43.960
<v Speaker 2>if you if you, if you think about the messages

0:03:44.000 --> 0:03:46.200
<v Speaker 2>that I was trying to communicate.

0:03:47.840 --> 0:03:49.000
<v Speaker 3>During the earning.

0:03:48.760 --> 0:03:54.040
<v Speaker 2>Stroll yesterday, I talked about the incredible h impact of

0:03:54.160 --> 0:03:57.400
<v Speaker 2>the effort that we've had to diversify our clients. We

0:03:57.440 --> 0:04:00.360
<v Speaker 2>started the year with you know, eighty five percent of

0:04:00.400 --> 0:04:03.560
<v Speaker 2>our revenue to a single client, and we are coming

0:04:03.640 --> 0:04:06.960
<v Speaker 2>out of the year with our back log with you know,

0:04:07.000 --> 0:04:09.880
<v Speaker 2>which has reached you know, all time highs where no

0:04:10.560 --> 0:04:14.120
<v Speaker 2>specific client is more than thirty five percent of it,

0:04:14.320 --> 0:04:16.520
<v Speaker 2>and that is down from fifty percent even last quarter.

0:04:16.560 --> 0:04:19.880
<v Speaker 2>So you're seeing tremendous progress on the contracting side. On

0:04:19.960 --> 0:04:23.640
<v Speaker 2>the supply side, when you're looking at the powershells the

0:04:23.720 --> 0:04:28.040
<v Speaker 2>data center providers, you know, there is no single provider

0:04:28.080 --> 0:04:33.520
<v Speaker 2>of data centers that represents in excess of twenty percent

0:04:33.760 --> 0:04:37.400
<v Speaker 2>of our two point nine gigawatts of power that we

0:04:37.440 --> 0:04:41.119
<v Speaker 2>will be delivering to the market over the next twenty

0:04:41.120 --> 0:04:41.880
<v Speaker 2>four months.

0:04:42.320 --> 0:04:42.560
<v Speaker 3>You know.

0:04:42.640 --> 0:04:45.919
<v Speaker 2>So you're seeing a real focus by our company to

0:04:46.200 --> 0:04:50.920
<v Speaker 2>diversify on both sides of the you know, the ledger,

0:04:52.080 --> 0:04:56.960
<v Speaker 2>the clients that we are delivering infrastructure to as well

0:04:57.000 --> 0:05:00.640
<v Speaker 2>as the suppliers that are delivering us the components that

0:05:00.680 --> 0:05:02.000
<v Speaker 2>we need to be successful.

0:05:02.480 --> 0:05:05.560
<v Speaker 1>What can be done Because we hear about power issues,

0:05:05.560 --> 0:05:07.960
<v Speaker 1>we hear about how much people maybe need government to

0:05:08.000 --> 0:05:11.760
<v Speaker 1>help speed up contracts, supply chain headaches. What is it

0:05:11.800 --> 0:05:15.040
<v Speaker 1>in this particular data center issue that won't be replicated

0:05:15.080 --> 0:05:16.919
<v Speaker 1>down the road? How are you making sure or is

0:05:16.920 --> 0:05:18.599
<v Speaker 1>it just something we are going to see time and

0:05:18.640 --> 0:05:20.000
<v Speaker 1>time again do you think so?

0:05:22.120 --> 0:05:25.080
<v Speaker 2>Look the way that we are handling this is we

0:05:25.120 --> 0:05:30.840
<v Speaker 2>are doing several discrete things to ensure that we are

0:05:32.080 --> 0:05:35.240
<v Speaker 2>future proofing ourselves against these type of delays. Right, So

0:05:36.400 --> 0:05:39.600
<v Speaker 2>we have our team that has been built out over

0:05:39.640 --> 0:05:43.600
<v Speaker 2>the past year that builds and delivers data center and

0:05:43.640 --> 0:05:46.480
<v Speaker 2>so we have a self build organization to be able

0:05:46.520 --> 0:05:49.720
<v Speaker 2>to deliver our own data centers for instance Kenilworth in

0:05:49.760 --> 0:05:52.960
<v Speaker 2>New Jersey or Lancaster in Pennsylvania, where we are building

0:05:53.560 --> 0:05:56.960
<v Speaker 2>from the ground up those skill sets, being able to

0:05:56.960 --> 0:06:01.559
<v Speaker 2>deploy those individuals with those talents down to the data

0:06:01.600 --> 0:06:06.440
<v Speaker 2>centers as they're being constructed by third parties. If they

0:06:07.160 --> 0:06:11.599
<v Speaker 2>encounter problems where we can be useful, helpful thinking through

0:06:11.680 --> 0:06:15.320
<v Speaker 2>ways of solving and driving the process forward is really

0:06:15.360 --> 0:06:18.400
<v Speaker 2>important to being able to mitigate these type of delays.

0:06:19.320 --> 0:06:21.720
<v Speaker 3>We have massively.

0:06:22.760 --> 0:06:26.880
<v Speaker 2>Diversified our data center providers so we're not exposed to

0:06:27.000 --> 0:06:30.080
<v Speaker 2>any specific they say to data center provider.

0:06:31.839 --> 0:06:34.200
<v Speaker 3>In too large of a component.

0:06:34.240 --> 0:06:36.920
<v Speaker 2>And then the third piece of it is it's important

0:06:36.920 --> 0:06:41.440
<v Speaker 2>to understand that as we get larger, as our delivered

0:06:41.480 --> 0:06:45.360
<v Speaker 2>power gets larger, the relative impact of a delay at

0:06:45.440 --> 0:06:49.760
<v Speaker 2>any one given site becomes less and less meaningful and

0:06:49.839 --> 0:06:52.560
<v Speaker 2>less and less impactful on our.

0:06:54.560 --> 0:06:56.599
<v Speaker 3>Run rate. And that's really important.

0:06:56.720 --> 0:07:02.080
<v Speaker 2>Like you're encountering scaling, is use within a company that's

0:07:02.279 --> 0:07:04.480
<v Speaker 2>encountering scaling issues within.

0:07:04.200 --> 0:07:07.440
<v Speaker 3>A supply chain, both of those will get better. Time

0:07:07.560 --> 0:07:10.280
<v Speaker 3>solves that scale solves that we.

0:07:10.200 --> 0:07:13.400
<v Speaker 1>Are currently seeing that time isn't particularly helping the share price.

0:07:13.440 --> 0:07:15.480
<v Speaker 1>As we speak, it's now down fourteen point three percent,

0:07:15.520 --> 0:07:17.680
<v Speaker 1>worst day since August. We are here with Michael and

0:07:17.720 --> 0:07:20.760
<v Speaker 1>trades across radio and TV, and I want to sort

0:07:20.760 --> 0:07:24.760
<v Speaker 1>of ask whether you think, more broadly the US is

0:07:24.800 --> 0:07:28.080
<v Speaker 1>stunted by the supply chain headache, or do you think

0:07:28.120 --> 0:07:30.160
<v Speaker 1>there is something that can be done from a federal

0:07:30.200 --> 0:07:32.480
<v Speaker 1>perspective here to ensure that you can build at the

0:07:32.560 --> 0:07:34.520
<v Speaker 1>rate you want to build. You're not having these sorts

0:07:34.520 --> 0:07:35.080
<v Speaker 1>of headaches.

0:07:35.480 --> 0:07:41.080
<v Speaker 2>So we've been we and many others have been, you know,

0:07:41.240 --> 0:07:45.680
<v Speaker 2>quite forceful in making the case that there is a

0:07:45.800 --> 0:07:50.560
<v Speaker 2>role for government in helping us with some of the

0:07:50.600 --> 0:07:55.720
<v Speaker 2>permitting issues, speed to which you can get our infrastructure

0:07:56.320 --> 0:07:58.760
<v Speaker 2>and others can get their infrastructure attached to the grid,

0:07:58.960 --> 0:08:01.560
<v Speaker 2>you know, all those types of things. Is where there's

0:08:01.680 --> 0:08:04.480
<v Speaker 2>there's there's a great role for for for the government

0:08:04.680 --> 0:08:07.560
<v Speaker 2>to help in that. They are the they are the

0:08:07.720 --> 0:08:12.040
<v Speaker 2>organization or government entity that is correctly positioned to help

0:08:12.200 --> 0:08:15.360
<v Speaker 2>facilitate that. And I think that's a great role where

0:08:15.440 --> 0:08:18.560
<v Speaker 2>where government can lean in and make an impact across

0:08:18.600 --> 0:08:19.000
<v Speaker 2>the space.

0:08:19.120 --> 0:08:23.400
<v Speaker 4>Rob Michael, you have an agreement within Nvidia that lots

0:08:23.440 --> 0:08:25.840
<v Speaker 4>of people find very interesting that down the road, if

0:08:25.880 --> 0:08:30.120
<v Speaker 4>there is spare capacity, there's some flexibility for you to

0:08:30.200 --> 0:08:33.840
<v Speaker 4>deploy it elsewhere. That's a simple summation of it. But

0:08:34.160 --> 0:08:36.440
<v Speaker 4>I'm wondering how you're thinking about serving some of the

0:08:36.480 --> 0:08:40.560
<v Speaker 4>smaller AI labs and casting that net even wider in

0:08:40.600 --> 0:08:44.440
<v Speaker 4>your customer base if indeed that capacity gets freed up

0:08:44.440 --> 0:08:45.160
<v Speaker 4>down the line.

0:08:45.679 --> 0:08:49.200
<v Speaker 2>Yeah, it's a it's a contract I'm actually incredibly excited about. Right,

0:08:49.280 --> 0:08:52.840
<v Speaker 2>It's a contract that we did within Video where we

0:08:52.920 --> 0:08:56.280
<v Speaker 2>will deliver them compute for the next six years. But

0:08:56.960 --> 0:08:59.920
<v Speaker 2>in the contract is the capacity to interrupt the flow

0:09:00.000 --> 0:09:03.960
<v Speaker 2>of compute to them and that will allow us to

0:09:04.320 --> 0:09:10.720
<v Speaker 2>repurpose that compute to new companies, startups, companies that have

0:09:10.880 --> 0:09:13.880
<v Speaker 2>struggled to get access to the compute that they require

0:09:14.520 --> 0:09:19.560
<v Speaker 2>to bring new companies, organizations, ideas into existence. And I

0:09:19.559 --> 0:09:25.280
<v Speaker 2>think it's just an incredibly important component of how the

0:09:25.320 --> 0:09:29.679
<v Speaker 2>infrastructure is so important to allowing the ecosystem itself of

0:09:29.760 --> 0:09:32.359
<v Speaker 2>these startups, of these new companies, of these new ideas

0:09:32.480 --> 0:09:38.000
<v Speaker 2>to become more resilient, to become more scaled, and it's

0:09:38.040 --> 0:09:39.640
<v Speaker 2>just a great contract for us to.

0:09:39.559 --> 0:09:41.000
<v Speaker 3>Be able to position ourselves.

0:09:41.120 --> 0:09:44.600
<v Speaker 2>It also provides this wonderful on ramp for us to

0:09:44.600 --> 0:09:48.040
<v Speaker 2>be able to work with the new amazing companies that

0:09:48.080 --> 0:09:51.240
<v Speaker 2>are coming into existence so that they integrate into our

0:09:51.400 --> 0:09:54.560
<v Speaker 2>solution and get to make use of the best alternative

0:09:54.600 --> 0:09:57.000
<v Speaker 2>that exists in the market as they're scaling their companies.

0:09:58.120 --> 0:10:01.200
<v Speaker 4>Michael, it's finished to what extent is in Vidious still

0:10:01.240 --> 0:10:04.319
<v Speaker 4>the gold standard in videous GPUs for your customer base

0:10:04.760 --> 0:10:07.360
<v Speaker 4>and what data are you tracking on demand for those

0:10:07.440 --> 0:10:12.680
<v Speaker 4>kind of more inferenced specific chips that are offered by others.

0:10:13.400 --> 0:10:18.840
<v Speaker 2>So, look, we have always been client led, right. Our

0:10:18.960 --> 0:10:22.280
<v Speaker 2>clients come to market and tell us, hey, we would

0:10:22.320 --> 0:10:25.600
<v Speaker 2>like you guys to help us build a cluster. We

0:10:25.679 --> 0:10:27.840
<v Speaker 2>needed to be this size and this location and this

0:10:27.880 --> 0:10:30.160
<v Speaker 2>type of network, and we work with them in a

0:10:30.280 --> 0:10:34.560
<v Speaker 2>very kind of interactive way to ensure that the infrastructure

0:10:34.600 --> 0:10:38.040
<v Speaker 2>that we're building is the best infrastructure for them, is

0:10:38.120 --> 0:10:41.920
<v Speaker 2>fungible for us, like all of these type of requirements

0:10:42.000 --> 0:10:46.079
<v Speaker 2>to make for successful delivery of infrastructure. Right now, the

0:10:47.000 --> 0:10:52.319
<v Speaker 2>reality of the situation is the buy signals from our

0:10:52.559 --> 0:10:58.280
<v Speaker 2>clients overwhelm our capacity to deliver infrastructure to the markets.

0:10:58.280 --> 0:11:02.360
<v Speaker 2>Matter of fact, they overwhelmed hire markets capacity to deliver

0:11:02.559 --> 0:11:06.800
<v Speaker 2>infrastructure to the market. You have a systemic shortage of

0:11:08.080 --> 0:11:12.800
<v Speaker 2>ability to deliver the GPUs the computing infrastructure for the

0:11:12.840 --> 0:11:16.320
<v Speaker 2>buildout of artificial intelligence. And we have never wavered from

0:11:16.320 --> 0:11:19.800
<v Speaker 2>that position. We have been very very clear that when

0:11:19.840 --> 0:11:22.520
<v Speaker 2>we look at the demand signals coming into core Weave,

0:11:22.559 --> 0:11:26.040
<v Speaker 2>the totality of that overwhelms the capacity of the market

0:11:26.080 --> 0:11:28.719
<v Speaker 2>to deliver that, and we'll continue to do that for.

0:11:28.720 --> 0:11:29.280
<v Speaker 3>Quite a while.

0:11:30.720 --> 0:11:33.320
<v Speaker 4>Michael and tracer Core. We've CEO, thank you for coming

0:11:33.360 --> 0:11:34.680
<v Speaker 4>back on Bloomberg Tech