1
00:00:02,720 --> 00:00:15,800
Speaker 1: Bloomberg Audio Studios, Podcasts, radio News.

2
00:00:18,600 --> 00:00:21,640
Speaker 2: Hello and welcome to another episode of the Odd Lots podcast.

3
00:00:21,720 --> 00:00:23,919
Speaker 3: I'm Jill Wisenthal and I'm Tracy Alloway.

4
00:00:24,200 --> 00:00:26,360
Speaker 2: Tracy, you know, we've obviously been doing a lot about

5
00:00:26,480 --> 00:00:28,960
Speaker 2: data centers, but one thing that's sort of becoming interesting

6
00:00:29,000 --> 00:00:31,000
Speaker 2: is they just seem so complicated. There are so many

7
00:00:31,000 --> 00:00:34,120
Speaker 2: different moving parts. And I don't mean like technically complicated,

8
00:00:34,159 --> 00:00:36,400
Speaker 2: although that's part of it, but they really are this

9
00:00:36,520 --> 00:00:39,080
Speaker 2: sort of weird thing where they're a high tech thing.

10
00:00:39,280 --> 00:00:41,320
Speaker 2: They have the most advanced chips, or at least some

11
00:00:41,360 --> 00:00:42,680
Speaker 2: of them have the most advanced.

12
00:00:42,360 --> 00:00:43,080
Speaker 4: Chips in the world.

13
00:00:43,560 --> 00:00:46,720
Speaker 2: They're real estate plays and we talk about securitization and

14
00:00:46,760 --> 00:00:48,919
Speaker 2: so forth, and then they have to figure out all

15
00:00:48,920 --> 00:00:50,720
Speaker 2: this stuff of how they're going to connect to the

16
00:00:50,720 --> 00:00:52,840
Speaker 2: grid and maybe they're going to build their own power

17
00:00:52,880 --> 00:00:56,000
Speaker 2: plant inside the data center, et cetera. There's a lot

18
00:00:56,040 --> 00:00:56,760
Speaker 2: of moving parts.

19
00:00:57,040 --> 00:00:59,319
Speaker 3: Yeah, there's a couple of things here. So number one,

20
00:00:59,360 --> 00:01:01,880
Speaker 3: it does seem to be a complex space. And whenever

21
00:01:01,920 --> 00:01:04,800
Speaker 3: I think about how some of these things are being financed,

22
00:01:04,840 --> 00:01:07,400
Speaker 3: I get that, what's the meme with the guy standing

23
00:01:07,440 --> 00:01:09,520
Speaker 3: in front of the board with like all the papers

24
00:01:09,600 --> 00:01:13,640
Speaker 3: and directions. It all feels a little circular sometimes that

25
00:01:13,680 --> 00:01:16,800
Speaker 3: whole ecosystem. But the other thing is just the sheer

26
00:01:16,959 --> 00:01:21,039
Speaker 3: scale of the like financing requirement for doing this. So

27
00:01:21,080 --> 00:01:22,959
Speaker 3: I was reading a Morgan Stanley report. I think it

28
00:01:23,000 --> 00:01:26,119
Speaker 3: came out over the summer. They were forecasting two point

29
00:01:26,280 --> 00:01:31,479
Speaker 3: nine trillion of global data centers spend through twenty twenty eight.

30
00:01:31,640 --> 00:01:35,360
Speaker 3: And just for context, total CAPEC spending by all companies

31
00:01:35,400 --> 00:01:37,560
Speaker 3: in the S and P five hundred in twenty twenty

32
00:01:37,600 --> 00:01:39,080
Speaker 3: four was nine hundred and fifty billion.

33
00:01:39,160 --> 00:01:42,000
Speaker 2: It's crazy, yeah, And by the words, are the numbers bug?

34
00:01:42,280 --> 00:01:44,280
Speaker 3: Well? The other big question is if you actually stack

35
00:01:44,360 --> 00:01:48,600
Speaker 3: that up against existing revenue from generative AI, existing revenues

36
00:01:48,640 --> 00:01:51,520
Speaker 3: like sixteen billion, right, So there's a bit of a

37
00:01:51,560 --> 00:01:53,720
Speaker 3: gap there that needs to be filled.

38
00:01:53,480 --> 00:01:55,840
Speaker 2: Right, completely correct. The other thing that strikes me is

39
00:01:55,920 --> 00:01:59,200
Speaker 2: very interesting and tricky is you have these projects and

40
00:01:59,680 --> 00:02:02,960
Speaker 2: put there's big capital outlay, capital commitments, et cetera. And

41
00:02:03,000 --> 00:02:05,520
Speaker 2: then the timelines like oh, when do you get connected

42
00:02:05,560 --> 00:02:08,760
Speaker 2: to the grid, When does that turbine for your insight

43
00:02:08,880 --> 00:02:12,560
Speaker 2: gas generator get actually get delivered? Maybe twenty thirty et cetera.

44
00:02:12,600 --> 00:02:15,079
Speaker 2: So it feels like, I get the pressure. You're dealing

45
00:02:15,160 --> 00:02:18,480
Speaker 2: with very uncertain timelines, and so how you establish that

46
00:02:18,520 --> 00:02:21,000
Speaker 2: in the language so everyone feels protected, Like, what if

47
00:02:21,040 --> 00:02:23,560
Speaker 2: you could ever get connected, what if anything could happen?

48
00:02:23,639 --> 00:02:23,760
Speaker 5: Right?

49
00:02:23,760 --> 00:02:26,200
Speaker 3: What are the expectations for when this actually starts to

50
00:02:26,240 --> 00:02:26,800
Speaker 3: pay off.

51
00:02:26,639 --> 00:02:29,280
Speaker 2: When does it payoff, what are the obligations of the tenant,

52
00:02:29,440 --> 00:02:32,400
Speaker 2: et cetera. We don't really know anyway, I'm sure it's

53
00:02:32,400 --> 00:02:35,440
Speaker 2: different number one, So many more questions for just scratching

54
00:02:35,480 --> 00:02:35,880
Speaker 2: the surface.

55
00:02:36,080 --> 00:02:38,639
Speaker 3: Yeah, absolutely, And you know, there have been some interesting

56
00:02:38,639 --> 00:02:41,160
Speaker 3: things happening. So a lot of the activity has been

57
00:02:41,200 --> 00:02:45,040
Speaker 3: done in the private market. But obviously AI players and

58
00:02:45,120 --> 00:02:47,760
Speaker 3: companies have a lot of different options. So they could

59
00:02:47,760 --> 00:02:50,480
Speaker 3: go to securitized markets, and I think we've been seeing

60
00:02:50,520 --> 00:02:53,919
Speaker 3: some CMBs and ABS deals there, private credit I mentioned

61
00:02:53,919 --> 00:02:56,960
Speaker 3: private equity bank lending, or they can go the public

62
00:02:57,080 --> 00:03:00,359
Speaker 3: route into corporate bonds, and so I'm really curious how

63
00:03:00,480 --> 00:03:03,160
Speaker 3: a lot of these players are choosing between those options.

64
00:03:03,320 --> 00:03:05,760
Speaker 2: Well, I'm really excited to say we have the perfect guest,

65
00:03:05,880 --> 00:03:08,720
Speaker 2: someone who is really at the intersection of sort of

66
00:03:08,800 --> 00:03:12,000
Speaker 2: every one of the things that these data centers are

67
00:03:12,040 --> 00:03:14,040
Speaker 2: also at the intersection We're going to be speaking to

68
00:03:14,080 --> 00:03:16,880
Speaker 2: Travis Wafford. He's a partner at the law firm Baker Bods.

69
00:03:16,919 --> 00:03:20,280
Speaker 2: He is the chair of the corporate department based in Houston.

70
00:03:20,440 --> 00:03:22,440
Speaker 2: He works on all of these things. He can tell

71
00:03:22,520 --> 00:03:24,920
Speaker 2: us what's in the details of all these agreements. So, Travis,

72
00:03:24,960 --> 00:03:27,240
Speaker 2: thank you so much for coming on. Odd Love's great

73
00:03:27,280 --> 00:03:28,120
Speaker 2: to have you here in studio.

74
00:03:28,160 --> 00:03:28,880
Speaker 4: Thanks for having me.

75
00:03:29,040 --> 00:03:31,480
Speaker 2: What do you give us just a quick overview? What's

76
00:03:31,480 --> 00:03:33,880
Speaker 2: your role at Baker Boss? What do you focus on?

77
00:03:33,919 --> 00:03:35,800
Speaker 2: What do you call it a practice area? Is that

78
00:03:35,880 --> 00:03:38,200
Speaker 2: a legal for a legal term?

79
00:03:38,320 --> 00:03:42,440
Speaker 5: Yeah, So I'm a deal lawyer. I help people build, buy, finance,

80
00:03:42,520 --> 00:03:46,520
Speaker 5: and sell projects and companies. And so we take multi

81
00:03:46,560 --> 00:03:49,320
Speaker 5: disciplinary teams and put them together in order to actually

82
00:03:49,720 --> 00:03:51,880
Speaker 5: make people's dreams. The reality is the way I like

83
00:03:51,960 --> 00:03:55,360
Speaker 5: to explain it. The idea is somebody comes to us

84
00:03:55,400 --> 00:03:58,840
Speaker 5: with a very complex project and they need an integrated solution,

85
00:03:59,120 --> 00:04:02,360
Speaker 5: legal solution to that problem. So you want to build

86
00:04:02,360 --> 00:04:04,720
Speaker 5: the data center, you've got power, you need people that

87
00:04:04,760 --> 00:04:09,720
Speaker 5: actually understand the air, the water, the interconnection with the fiber,

88
00:04:10,160 --> 00:04:13,760
Speaker 5: the interconnection with power, the land, everything else. And then

89
00:04:13,800 --> 00:04:16,320
Speaker 5: the finance layer on top of that we provide a

90
00:04:16,320 --> 00:04:17,960
Speaker 5: one stop shop for that.

91
00:04:18,400 --> 00:04:21,479
Speaker 3: Okay, so my question is how busy are you right now?

92
00:04:21,520 --> 00:04:25,039
Speaker 3: Are you just getting you know, requests flying at you constantly?

93
00:04:25,360 --> 00:04:28,440
Speaker 5: Yes, we are very busy, which is a good thing.

94
00:04:28,480 --> 00:04:31,200
Speaker 5: And it's a bad thing, right because your capacity constrain,

95
00:04:31,320 --> 00:04:32,320
Speaker 5: just like your clients are.

96
00:04:32,400 --> 00:04:34,640
Speaker 3: Well, my husband is a former lawyer, so I know

97
00:04:34,720 --> 00:04:35,719
Speaker 3: how bad it can be.

98
00:04:36,040 --> 00:04:39,000
Speaker 5: Yes, But the good thing is they're interesting deals. It's

99
00:04:39,080 --> 00:04:41,279
Speaker 5: interesting stuff to work on. It's in the news on

100
00:04:41,600 --> 00:04:44,120
Speaker 5: Bloomberg as well, and that makes it fun.

101
00:04:44,400 --> 00:04:46,920
Speaker 2: It seems like a baker Bod's with like the long

102
00:04:47,120 --> 00:04:50,440
Speaker 2: history based in Houston, all the energy stuff. I mean,

103
00:04:50,720 --> 00:04:53,840
Speaker 2: this is all sort of novel to everyone, right, especially

104
00:04:54,040 --> 00:04:57,120
Speaker 2: the generative AI data center Like that aspect is just

105
00:04:57,160 --> 00:04:59,599
Speaker 2: a couple of years old. Can you talk about when

106
00:04:59,640 --> 00:05:02,200
Speaker 2: there is a new thing that everyone gets excited about

107
00:05:02,240 --> 00:05:04,320
Speaker 2: that maybe even a couple of years ago no one

108
00:05:04,360 --> 00:05:07,880
Speaker 2: was talking about now, granted, I know data centers a

109
00:05:07,880 --> 00:05:09,400
Speaker 2: lot longer than a couple of years old, but the

110
00:05:09,480 --> 00:05:12,479
Speaker 2: explosion of interest, how do you build that sort of

111
00:05:12,520 --> 00:05:15,800
Speaker 2: team that understands all of the dimensions so that you

112
00:05:15,839 --> 00:05:17,880
Speaker 2: could provide that one stop shop service.

113
00:05:18,000 --> 00:05:20,719
Speaker 5: Yeah, so it's a really good question. And you know,

114
00:05:20,800 --> 00:05:22,720
Speaker 5: I always go back to the expression there's nothing new

115
00:05:22,800 --> 00:05:25,479
Speaker 5: under the sun. When you actually think about what we're doing.

116
00:05:25,880 --> 00:05:29,160
Speaker 5: You're building a power project, you're building out a data center,

117
00:05:29,279 --> 00:05:31,400
Speaker 5: the powered shell or the powered land or the like.

118
00:05:31,839 --> 00:05:33,400
Speaker 4: People have been doing that before.

119
00:05:33,520 --> 00:05:35,600
Speaker 5: When you're financing these we're actually using a lot of

120
00:05:35,640 --> 00:05:38,960
Speaker 5: the same financing structures. It may not be exactly the same,

121
00:05:39,000 --> 00:05:42,400
Speaker 5: but it rhymes. So ten twenty years ago we were

122
00:05:42,400 --> 00:05:46,520
Speaker 5: doing whole business securitizations of cell towers, that same securitization

123
00:05:46,680 --> 00:05:50,880
Speaker 5: technology you moved forward to residential solar, and now you're

124
00:05:50,920 --> 00:05:55,000
Speaker 5: moving that same thing forward to CNBS and ABS of

125
00:05:55,040 --> 00:05:57,960
Speaker 5: the data centers. So they're very similar. The rating agencies

126
00:05:57,960 --> 00:06:00,839
Speaker 5: are very familiar with these things. You have to look

127
00:06:00,880 --> 00:06:03,800
Speaker 5: at a few different variables and understand them slightly differently,

128
00:06:03,839 --> 00:06:09,160
Speaker 5: but particularly energy infrastructure and telecom infrastructure. Those are two

129
00:06:09,200 --> 00:06:11,159
Speaker 5: things that we've been doing for decades.

130
00:06:11,520 --> 00:06:14,760
Speaker 3: Oh, walk us through the differences then between say, you

131
00:06:14,760 --> 00:06:17,359
Speaker 3: know a cell tower ABS or I don't know, a

132
00:06:17,400 --> 00:06:21,040
Speaker 3: solar power ABS versus a data center ABS or CMBs.

133
00:06:21,279 --> 00:06:24,080
Speaker 5: Sure, so a cell tower ABS. Really when you're looking

134
00:06:24,080 --> 00:06:27,080
Speaker 5: at that, you just have the tower. It's on land.

135
00:06:27,120 --> 00:06:29,320
Speaker 5: You may have somebody that's coming along in order to

136
00:06:29,320 --> 00:06:32,640
Speaker 5: mow the lawn around it, but they don't have much

137
00:06:32,760 --> 00:06:35,320
Speaker 5: in terms of the actual equipment that's on top of

138
00:06:35,360 --> 00:06:38,920
Speaker 5: the cell tower. When you have a residential solar securitization,

139
00:06:39,400 --> 00:06:42,279
Speaker 5: you put the solar panel on top of somebody's roof.

140
00:06:42,640 --> 00:06:45,719
Speaker 5: You're contracting the cash flows off of that. Again, on

141
00:06:45,800 --> 00:06:49,400
Speaker 5: the cell tower, you're contracting the cash flows off of that.

142
00:06:49,520 --> 00:06:51,800
Speaker 5: You pull those together and then you put a bond

143
00:06:51,880 --> 00:06:55,280
Speaker 5: on top of it. With a data center securitization, what

144
00:06:55,320 --> 00:06:58,200
Speaker 5: you're doing there is you have the cash flow from

145
00:06:58,720 --> 00:07:01,200
Speaker 5: the data center lease. You have the cash flow from

146
00:07:01,520 --> 00:07:03,480
Speaker 5: other aspects of the business that may be going along

147
00:07:03,520 --> 00:07:05,320
Speaker 5: with that, like related to the power and some of

148
00:07:05,320 --> 00:07:06,760
Speaker 5: the additional services.

149
00:07:07,240 --> 00:07:09,840
Speaker 4: You pool those. You have long term contracted.

150
00:07:09,360 --> 00:07:11,880
Speaker 5: Cash flows, and then a rating agency can rate those

151
00:07:11,920 --> 00:07:14,120
Speaker 5: as well. But at the end of the day, you're

152
00:07:14,160 --> 00:07:16,720
Speaker 5: trying to make sure that you have a special purpose

153
00:07:16,800 --> 00:07:19,680
Speaker 5: vehicle that has all the assets that it needs within it.

154
00:07:20,160 --> 00:07:23,240
Speaker 5: You have an separate entity that's handling the billing and

155
00:07:23,280 --> 00:07:26,880
Speaker 5: the collection, an entity that's handling the operation and maintenance

156
00:07:27,320 --> 00:07:30,520
Speaker 5: of the asset, and that's supposed to be a standalone

157
00:07:30,680 --> 00:07:33,920
Speaker 5: product that can move at least the five to seven

158
00:07:34,080 --> 00:07:37,440
Speaker 5: years until you get to the anticipated payment date, and

159
00:07:37,920 --> 00:07:41,520
Speaker 5: often it's about a thirty year rated final maturity.

160
00:07:42,160 --> 00:07:44,360
Speaker 3: So I have a question, and it's come up a

161
00:07:44,360 --> 00:07:47,400
Speaker 3: couple times on previous podcasts, but you mentioned, you know,

162
00:07:47,480 --> 00:07:50,640
Speaker 3: having a special purpose vehicle for this. Sure, Why does

163
00:07:50,680 --> 00:07:54,480
Speaker 3: it seem that so many big tech companies who are

164
00:07:54,480 --> 00:07:57,280
Speaker 3: in the AI space, who are presumably very, very cash

165
00:07:57,400 --> 00:08:00,960
Speaker 3: rich and profitable, why are so many of them financing

166
00:08:01,280 --> 00:08:02,239
Speaker 3: off balance sheet?

167
00:08:02,840 --> 00:08:05,160
Speaker 5: They have better things to do with their money, so

168
00:08:05,440 --> 00:08:09,280
Speaker 5: the return on their invested capital is much better on

169
00:08:09,320 --> 00:08:12,480
Speaker 5: the tech side than it is on the infrastructure side.

170
00:08:12,720 --> 00:08:17,280
Speaker 5: Infrastructure is usually low margins, but the benefit with that

171
00:08:17,480 --> 00:08:23,559
Speaker 5: is it's long term, reliable, understood assets, such as again,

172
00:08:23,600 --> 00:08:26,560
Speaker 5: if you're doing a large energy project, or even the

173
00:08:26,600 --> 00:08:30,360
Speaker 5: poweredshell itself. It's land, it's a building, it's got the

174
00:08:30,360 --> 00:08:33,720
Speaker 5: fiber connection, it's got the power connection, and that's something

175
00:08:33,720 --> 00:08:36,680
Speaker 5: that can continue over a long period of time, Whereas

176
00:08:36,720 --> 00:08:39,280
Speaker 5: on the tech side, they make a lot more money

177
00:08:39,280 --> 00:08:41,000
Speaker 5: and have much greater margins, as we all.

178
00:08:40,920 --> 00:08:44,000
Speaker 2: Know, right, So this aspect of the business, which could

179
00:08:44,000 --> 00:08:48,480
Speaker 2: be very profitable, but it's stable, it's very predictable. It's like,

180
00:08:48,559 --> 00:08:52,440
Speaker 2: let's outsource this, Let's have some other entity house that risk.

181
00:08:52,679 --> 00:08:55,360
Speaker 2: What are the risks in data center finance when you

182
00:08:55,360 --> 00:08:57,200
Speaker 2: think about the other side and not the tech companies,

183
00:08:57,240 --> 00:08:59,720
Speaker 2: but the lenders to the SPV and we'll talk about

184
00:08:59,720 --> 00:09:02,959
Speaker 2: thefferent structures that the credit form can take. What are

185
00:09:03,040 --> 00:09:05,480
Speaker 2: the risks? You know, it's a fairly stable, long term thing,

186
00:09:05,559 --> 00:09:06,960
Speaker 2: but how can things go wrong?

187
00:09:07,440 --> 00:09:10,000
Speaker 5: So it depends on what exactly you're financing, right, So,

188
00:09:10,040 --> 00:09:13,560
Speaker 5: in a securitization and ABS, what you're doing is you

189
00:09:13,600 --> 00:09:17,520
Speaker 5: are financing the business of this PowerShell or turnkey solution.

190
00:09:18,160 --> 00:09:21,840
Speaker 5: The tenant quality is very important. Do I have somebody

191
00:09:21,880 --> 00:09:26,079
Speaker 5: that has the investment grade tenant quality or diversified pool

192
00:09:26,120 --> 00:09:29,240
Speaker 5: of high credit quality tenants who can actually be clear?

193
00:09:29,320 --> 00:09:33,240
Speaker 3: You're securitizing the cash flows from the data center leases, right, yeah?

194
00:09:33,280 --> 00:09:34,000
Speaker 4: So exactly.

195
00:09:34,080 --> 00:09:38,240
Speaker 5: So on an ABS, it's a focus on the business

196
00:09:38,480 --> 00:09:42,800
Speaker 5: of this data center. The actual lease is in the securitization,

197
00:09:43,240 --> 00:09:45,600
Speaker 5: and then that the hard assets are in there as well,

198
00:09:45,920 --> 00:09:49,240
Speaker 5: and a CMBs you have the lease as well, but

199
00:09:49,280 --> 00:09:51,760
Speaker 5: it's more focused on the mortgage. And then we have

200
00:09:51,800 --> 00:09:52,600
Speaker 5: credit tenant.

201
00:09:52,440 --> 00:09:55,240
Speaker 4: Leases as well, which are kind of directly to the tenant.

202
00:09:55,280 --> 00:09:56,200
Speaker 3: That's useful, thank you.

203
00:09:56,720 --> 00:10:00,960
Speaker 5: So risk associated with securitization are usually focus on the

204
00:10:01,040 --> 00:10:04,199
Speaker 5: tenant quality. They're focus on the term of the lease.

205
00:10:04,559 --> 00:10:06,320
Speaker 5: How long is the lease if you have a ten

206
00:10:06,440 --> 00:10:09,240
Speaker 5: or fifteen year lease. If you have a shorter term lease,

207
00:10:09,520 --> 00:10:11,800
Speaker 5: you're going to look a little bit more at well,

208
00:10:12,480 --> 00:10:14,480
Speaker 5: is this something that's going to need to be released

209
00:10:14,760 --> 00:10:18,040
Speaker 5: in that timeframe? What is the rate on the lease?

210
00:10:18,160 --> 00:10:20,600
Speaker 5: If it's a below market lease, it's much less likely

211
00:10:20,640 --> 00:10:22,520
Speaker 5: that somebody is going to want to get a new

212
00:10:22,600 --> 00:10:26,760
Speaker 5: lease and leave you. And then the facility itself. Is

213
00:10:26,760 --> 00:10:30,120
Speaker 5: the facility going to be able to withstand kind of

214
00:10:30,160 --> 00:10:32,320
Speaker 5: technology risk over a long period of time?

215
00:10:32,600 --> 00:10:35,280
Speaker 3: So how are people structuring the deals at the moment

216
00:10:35,520 --> 00:10:38,959
Speaker 3: to compensate for that risk? So you have the time mismatch,

217
00:10:39,040 --> 00:10:43,160
Speaker 3: you have maybe tenancy rollover risk as well. Are people

218
00:10:43,200 --> 00:10:46,800
Speaker 3: like adding extra credit protection or how are they making

219
00:10:46,840 --> 00:10:48,400
Speaker 3: these palatable to investors?

220
00:10:48,880 --> 00:10:50,079
Speaker 4: So one is.

221
00:10:50,080 --> 00:10:54,040
Speaker 5: The loan to value your advance rate. So typically if

222
00:10:54,040 --> 00:10:57,240
Speaker 5: you're looking for something that's an investment grade securitization, you're

223
00:10:57,280 --> 00:11:01,040
Speaker 5: going to have it around forty to fifty percent. The

224
00:11:01,160 --> 00:11:04,959
Speaker 5: structure of the bond itself will have a rated final

225
00:11:05,000 --> 00:11:08,000
Speaker 5: maturity that's much further out than the lease, maybe twenty

226
00:11:08,000 --> 00:11:12,520
Speaker 5: five thirty years, but you're anticipating that the full principal

227
00:11:12,520 --> 00:11:14,400
Speaker 5: amount of the bond would be able to be repaid

228
00:11:14,600 --> 00:11:18,120
Speaker 5: within five to seven years. But the lease itself is

229
00:11:18,160 --> 00:11:21,000
Speaker 5: a ten year lease or a fifteen year lease, and

230
00:11:21,040 --> 00:11:23,960
Speaker 5: then that can get reapt for another ten years on

231
00:11:24,000 --> 00:11:27,720
Speaker 5: top of that. So when you're looking at your cash flows,

232
00:11:28,040 --> 00:11:32,680
Speaker 5: the cash flows are going to support your ard prior

233
00:11:32,760 --> 00:11:36,640
Speaker 5: to even worrying wait aird the anticipated repayment date of

234
00:11:36,679 --> 00:11:39,600
Speaker 5: the bond, prior to even worrying about your release.

235
00:11:40,000 --> 00:11:42,640
Speaker 2: So one of the things that sort of like people

236
00:11:42,679 --> 00:11:46,160
Speaker 2: tweet about and people even write about, including us, that

237
00:11:46,280 --> 00:11:49,720
Speaker 2: we've talked about even this whole thing about GPU life

238
00:11:49,760 --> 00:11:52,839
Speaker 2: and some of these questions regarding the long term value

239
00:11:52,880 --> 00:11:56,240
Speaker 2: of the assets, and people like you know, like to claim, oh,

240
00:11:56,320 --> 00:11:58,280
Speaker 2: they're not as going to be as valuable long term

241
00:11:58,320 --> 00:12:00,920
Speaker 2: as people think. Does this come up in your work

242
00:12:01,000 --> 00:12:02,600
Speaker 2: and what's really going on?

243
00:12:03,160 --> 00:12:06,320
Speaker 5: Sure, So it depends on what the financing structure is

244
00:12:06,400 --> 00:12:10,120
Speaker 5: that you're using. So the GPU life itself is more

245
00:12:10,240 --> 00:12:14,440
Speaker 5: question on a private credit story or an equity. On

246
00:12:14,480 --> 00:12:17,959
Speaker 5: the securitization side, they're more focused on the turnkey data

247
00:12:17,960 --> 00:12:21,480
Speaker 5: center itself. The useful life comes up a lot when

248
00:12:21,520 --> 00:12:23,960
Speaker 5: you're doing the accounting, and that rolls through to the

249
00:12:24,000 --> 00:12:26,280
Speaker 5: earnings per share of these public companies.

250
00:12:26,679 --> 00:12:28,679
Speaker 4: You've got a useful life of a.

251
00:12:28,640 --> 00:12:31,760
Speaker 5: Few years ago, it was three years for these GPUs.

252
00:12:32,200 --> 00:12:34,959
Speaker 5: Now some of them are using six years. If you're

253
00:12:34,960 --> 00:12:38,640
Speaker 5: doing straight line amortization over three years taken about a

254
00:12:38,640 --> 00:12:41,840
Speaker 5: third off over six years, it's obviously less than that,

255
00:12:42,160 --> 00:12:45,400
Speaker 5: and so you have much less amortization depreciation expense on it.

256
00:12:45,720 --> 00:12:47,320
Speaker 4: That helps your ranks for sure.

257
00:13:03,320 --> 00:13:06,240
Speaker 3: What's your impression of how private credit lenders have been

258
00:13:06,280 --> 00:13:10,400
Speaker 3: handling the GPU aging problem so far, Like, are they

259
00:13:10,440 --> 00:13:13,040
Speaker 3: conscious of it or are people still kind of basing

260
00:13:13,080 --> 00:13:16,880
Speaker 3: a lot of their expected cash flows off of GPUs

261
00:13:16,920 --> 00:13:19,480
Speaker 3: that last, you know, longer because they were previously being

262
00:13:19,559 --> 00:13:21,559
Speaker 3: used in cloud computing or something like that.

263
00:13:21,760 --> 00:13:25,160
Speaker 5: Yeah, so they're very aware of it, particularly because if

264
00:13:25,240 --> 00:13:27,880
Speaker 5: you look at the new in video had the H

265
00:13:27,960 --> 00:13:30,200
Speaker 5: one hundreds and the H two hundreds that had come out.

266
00:13:30,640 --> 00:13:33,920
Speaker 5: Then it was the Blackwell the B one hundreds two hundreds,

267
00:13:34,120 --> 00:13:36,600
Speaker 5: and now Reuben's going to come after that, right, So

268
00:13:36,760 --> 00:13:40,720
Speaker 5: every two to three years you have these new chipsets

269
00:13:41,040 --> 00:13:43,920
Speaker 5: and that it's almost like, do I want to be

270
00:13:44,280 --> 00:13:46,680
Speaker 5: financing these laptops that you've got in front of me

271
00:13:46,800 --> 00:13:50,840
Speaker 5: for six years or seven years? What's the replacement cycle

272
00:13:50,880 --> 00:13:52,760
Speaker 5: on that? So when do I need to get repaid

273
00:13:53,200 --> 00:13:57,360
Speaker 5: on my debt? The interesting part that comes into it

274
00:13:57,400 --> 00:14:00,839
Speaker 5: is it's not just one laptop. We're talking about entire

275
00:14:00,920 --> 00:14:04,600
Speaker 5: data centers full of these things, and that's in the

276
00:14:05,120 --> 00:14:08,120
Speaker 5: tens hundreds of millions of dollars. So how are you

277
00:14:08,200 --> 00:14:12,080
Speaker 5: actually going to get your cash flow back when maybe

278
00:14:12,080 --> 00:14:14,240
Speaker 5: those aren't going to be used for training the same way.

279
00:14:14,880 --> 00:14:18,480
Speaker 5: The reality though, is the useful life of these and

280
00:14:18,520 --> 00:14:23,240
Speaker 5: the economic life is not just that initial usage.

281
00:14:23,280 --> 00:14:25,040
Speaker 4: You can use it for other things.

282
00:14:25,120 --> 00:14:28,960
Speaker 5: So starting out with training, needing to get trained done

283
00:14:28,960 --> 00:14:32,040
Speaker 5: as quickly as possible. If the data center is well

284
00:14:32,080 --> 00:14:36,760
Speaker 5: located in Virginia or another Tier one location, maybe there's

285
00:14:36,880 --> 00:14:40,440
Speaker 5: inference if it's close to a city or other location.

286
00:14:40,960 --> 00:14:44,960
Speaker 5: And then for the CPUs data centers that have those,

287
00:14:45,040 --> 00:14:48,200
Speaker 5: you can use them for compute and analytics and other

288
00:14:48,840 --> 00:14:51,840
Speaker 5: technologies and use cases that maybe that one what you

289
00:14:51,960 --> 00:14:55,160
Speaker 5: originally underwrote, but you knew that that was coming down

290
00:14:55,200 --> 00:14:55,680
Speaker 5: the line.

291
00:14:56,000 --> 00:14:58,680
Speaker 2: Since you mentioned Virginia, can you talk about where we

292
00:14:58,720 --> 00:15:02,200
Speaker 2: are in twenty twenty five with data center siting and

293
00:15:02,360 --> 00:15:06,120
Speaker 2: picking locations. So I know that there's tons around northern Virginia,

294
00:15:06,520 --> 00:15:08,480
Speaker 2: and I know that there's some huge projects in the

295
00:15:08,480 --> 00:15:10,800
Speaker 2: middle of nowhere in Texas where they're not going to

296
00:15:10,800 --> 00:15:13,840
Speaker 2: offend any neighbors or whatever, etc. But what are the

297
00:15:13,920 --> 00:15:17,120
Speaker 2: big themes that companies are thinking about right now when

298
00:15:17,120 --> 00:15:18,440
Speaker 2: they think about location.

299
00:15:18,840 --> 00:15:21,760
Speaker 5: Yeah. So one of the great things about Virginia is

300
00:15:22,280 --> 00:15:26,080
Speaker 5: connectivity to the subse cables, so getting to Europe, getting

301
00:15:26,120 --> 00:15:29,480
Speaker 5: to Africa, getting to other locations and the rest of

302
00:15:29,520 --> 00:15:33,600
Speaker 5: the United States. There's a great regulatory environment there in

303
00:15:33,680 --> 00:15:37,960
Speaker 5: terms of power, in terms of actually building out the

304
00:15:38,040 --> 00:15:41,240
Speaker 5: infrastructure there and other data centers so that you can

305
00:15:41,360 --> 00:15:44,000
Speaker 5: have I hate to use the word co located, but

306
00:15:44,480 --> 00:15:48,080
Speaker 5: connectivity within other data centers makes it very attractive. The

307
00:15:48,160 --> 00:15:52,240
Speaker 5: problem is it's saturated. It's a very saturated market. And

308
00:15:52,320 --> 00:15:55,480
Speaker 5: so if you want to build that next large giant

309
00:15:55,560 --> 00:15:59,240
Speaker 5: data center and you need power, or you need water,

310
00:15:59,440 --> 00:16:01,960
Speaker 5: or you need other parts of the infrastructure, and you

311
00:16:02,000 --> 00:16:04,720
Speaker 5: need to do it quickly, You're often going to look elsewhere.

312
00:16:04,920 --> 00:16:06,960
Speaker 5: That's one of the reasons that Texas and ur God

313
00:16:07,000 --> 00:16:08,080
Speaker 5: have become so attractive.

314
00:16:08,400 --> 00:16:08,640
Speaker 1: Yeah.

315
00:16:08,640 --> 00:16:10,880
Speaker 3: So one of the things we hear constantly is that

316
00:16:11,120 --> 00:16:14,040
Speaker 3: chips and financing are not necessarily the big choke points

317
00:16:14,080 --> 00:16:16,960
Speaker 3: for doing this. It's more the power. And I guess

318
00:16:17,000 --> 00:16:19,680
Speaker 3: I'm curious, is it really a power shortage problem or

319
00:16:19,720 --> 00:16:22,960
Speaker 3: is it more the distribution of power is not in

320
00:16:23,000 --> 00:16:25,800
Speaker 3: the places where these data centers you know, want to

321
00:16:25,840 --> 00:16:29,440
Speaker 3: be because of convenience, like colocation. It's you can say

322
00:16:29,480 --> 00:16:30,160
Speaker 3: both as well.

323
00:16:30,480 --> 00:16:33,800
Speaker 5: Yeah, it definitely is both. I will say though that

324
00:16:33,880 --> 00:16:38,440
Speaker 5: with power its interconnection. It's getting that actual permission both

325
00:16:38,560 --> 00:16:42,120
Speaker 5: for the load, so the data center itself is drawing

326
00:16:42,160 --> 00:16:45,400
Speaker 5: power from the grid and for the generation and those

327
00:16:45,400 --> 00:16:49,280
Speaker 5: are two separate interconnections on these larger facilities that you're

328
00:16:49,320 --> 00:16:52,520
Speaker 5: going to be getting at relatively the same time. And

329
00:16:52,560 --> 00:16:54,800
Speaker 5: if you can't have both, then you're probably not going

330
00:16:54,840 --> 00:16:58,080
Speaker 5: to be able to support such a large facility. So

331
00:16:59,120 --> 00:17:02,640
Speaker 5: on the generationation side, if you're trying to get interconnection,

332
00:17:03,400 --> 00:17:07,200
Speaker 5: you go through a long process potentially with FURK and

333
00:17:07,280 --> 00:17:10,320
Speaker 5: with your ISO or rto you know, ERCOT if you're

334
00:17:10,359 --> 00:17:14,760
Speaker 5: in Texas to do typically at this scale, like the

335
00:17:14,920 --> 00:17:18,879
Speaker 5: large data center scale, like a five year process in

336
00:17:19,040 --> 00:17:23,000
Speaker 5: order to get that generation approved and interconnected, and.

337
00:17:22,920 --> 00:17:25,639
Speaker 3: It's that's just the approval five years and then you

338
00:17:25,680 --> 00:17:27,200
Speaker 3: have to actually do the connection.

339
00:17:27,359 --> 00:17:28,399
Speaker 4: Yeah, and then you got to build it.

340
00:17:28,440 --> 00:17:28,800
Speaker 2: Wow.

341
00:17:29,320 --> 00:17:31,960
Speaker 5: The fun part about that is that timeline. A lot

342
00:17:32,000 --> 00:17:34,000
Speaker 5: of people walk in and they think, oh, I've heard

343
00:17:34,000 --> 00:17:36,720
Speaker 5: that it's going to be eighteen to thirty six months

344
00:17:36,760 --> 00:17:39,440
Speaker 5: in order for me to get this done. And then

345
00:17:39,680 --> 00:17:41,800
Speaker 5: a month or two goes by and they get a

346
00:17:41,840 --> 00:17:44,000
Speaker 5: new date and it's been pushed out. And then a

347
00:17:44,000 --> 00:17:46,280
Speaker 5: few months go by and they get a new date

348
00:17:46,400 --> 00:17:49,080
Speaker 5: and it's been pushed out. And one of the analogies

349
00:17:49,080 --> 00:17:51,000
Speaker 5: we use is like you're waiting in the airport to

350
00:17:51,040 --> 00:17:52,120
Speaker 5: get on the plane.

351
00:17:52,280 --> 00:17:55,840
Speaker 4: And they call I fly United all the time.

352
00:17:55,840 --> 00:17:59,439
Speaker 5: They're like, oh, it's Global Services, okay, Global services, go forward,

353
00:17:59,760 --> 00:18:03,639
Speaker 5: and now armed services okay, armed services and people with

354
00:18:03,680 --> 00:18:06,600
Speaker 5: young children, and you just you realize that that first

355
00:18:06,600 --> 00:18:10,080
Speaker 5: class ticket that has you know, boarding group one, doesn't

356
00:18:10,160 --> 00:18:11,880
Speaker 5: mean that you're getting on first.

357
00:18:11,680 --> 00:18:13,080
Speaker 2: Well, you all get there at the same time.

358
00:18:13,080 --> 00:18:13,520
Speaker 4: At least.

359
00:18:13,520 --> 00:18:15,000
Speaker 2: This is how I cause you know, I don't like

360
00:18:15,040 --> 00:18:16,399
Speaker 2: waiting there either, but I was like, you know what

361
00:18:16,440 --> 00:18:18,480
Speaker 2: I'm getting. I'm going to arrive at the destination the

362
00:18:18,480 --> 00:18:21,400
Speaker 2: same as all these people with children everything, So that's

363
00:18:21,400 --> 00:18:24,080
Speaker 2: hot anyway. You know, there was a story recently Core

364
00:18:24,160 --> 00:18:26,600
Speaker 2: we have one of the big neocloud companies sort of

365
00:18:26,600 --> 00:18:30,920
Speaker 2: lowered a growth forecast because of delay and the third

366
00:18:31,119 --> 00:18:33,760
Speaker 2: party data center project, which is exactly sort of what

367
00:18:33,800 --> 00:18:35,960
Speaker 2: you're talking about these things. How does that affect the

368
00:18:36,000 --> 00:18:40,480
Speaker 2: financing this uncertainty of when you can actually plug these

369
00:18:40,480 --> 00:18:43,200
Speaker 2: things in or when they're going to get approved, because

370
00:18:43,720 --> 00:18:46,360
Speaker 2: that sounds very frustrating and time is money, and so

371
00:18:46,400 --> 00:18:48,840
Speaker 2: how does that interact with the credit component?

372
00:18:49,080 --> 00:18:51,919
Speaker 5: Yeah, so you kind of have three categories in the

373
00:18:51,960 --> 00:18:55,680
Speaker 5: timeline of one of these projects. You have your development

374
00:18:55,760 --> 00:19:00,080
Speaker 5: capital that's often equity funded. You have your construction but

375
00:19:00,160 --> 00:19:02,680
Speaker 5: also you get a big construction loan once you've got

376
00:19:02,680 --> 00:19:06,440
Speaker 5: your permits and your power, and then after that's been completed,

377
00:19:06,920 --> 00:19:10,680
Speaker 5: then you do your takeout finance. Okay, that maybe the securitization.

378
00:19:10,040 --> 00:19:10,480
Speaker 3: And the like.

379
00:19:10,920 --> 00:19:14,720
Speaker 2: Has the regulatory I guess political environment does it feel

380
00:19:14,760 --> 00:19:17,040
Speaker 2: like it's significantly changed in the last six months, So

381
00:19:17,080 --> 00:19:19,400
Speaker 2: you talk about like, you know, very few people were

382
00:19:19,440 --> 00:19:22,840
Speaker 2: talking about water. My impression is that the water component

383
00:19:22,880 --> 00:19:25,399
Speaker 2: specifically is very overrated based I think, except right, But

384
00:19:25,920 --> 00:19:28,159
Speaker 2: it obviously is a matter. But you know, the public

385
00:19:28,240 --> 00:19:31,040
Speaker 2: is really concerned about water. The public is clearly concerned

386
00:19:31,040 --> 00:19:33,959
Speaker 2: about electricity prices. People are showing up to town halls

387
00:19:34,000 --> 00:19:37,080
Speaker 2: to protest or to voice their opposition to new data

388
00:19:37,080 --> 00:19:39,679
Speaker 2: center projects in the areas where they live. Does it

389
00:19:39,720 --> 00:19:42,960
Speaker 2: feel like the environment today when we're talking in December

390
00:19:42,960 --> 00:19:45,320
Speaker 2: twenty twenty five, is meaningfully different than it was at

391
00:19:45,320 --> 00:19:47,800
Speaker 2: the start of the year in terms of public awareness

392
00:19:47,840 --> 00:19:48,479
Speaker 2: of all this stuff.

393
00:19:48,560 --> 00:19:53,040
Speaker 5: So politically, absolutely right. I think that the community organizers,

394
00:19:53,080 --> 00:19:55,760
Speaker 5: and there are a lot of organizations that kind of

395
00:19:55,840 --> 00:19:58,879
Speaker 5: make their money based on outrage around these types of things,

396
00:19:59,160 --> 00:20:00,880
Speaker 5: they've realized that this this is a great.

397
00:20:00,720 --> 00:20:01,760
Speaker 4: Opportunity for them.

398
00:20:01,800 --> 00:20:04,560
Speaker 5: And if you look at the sustainability reports of a

399
00:20:04,560 --> 00:20:08,080
Speaker 5: lot of the hyperscalers and others, they talk about being

400
00:20:08,160 --> 00:20:11,160
Speaker 5: water positive. It doesn't mean that they're just happy about water.

401
00:20:11,480 --> 00:20:13,800
Speaker 5: It means that they're actually trying to have a positive impact,

402
00:20:13,840 --> 00:20:16,680
Speaker 5: and they'll start putting money into the water sheds and

403
00:20:16,720 --> 00:20:21,000
Speaker 5: the like and water replacement and that actually is really meaningful.

404
00:20:21,160 --> 00:20:22,560
Speaker 5: Is a part of the story that I think a

405
00:20:22,560 --> 00:20:25,480
Speaker 5: lot of the folks that are thinking about the political

406
00:20:25,520 --> 00:20:28,320
Speaker 5: aspect of this don't realize, which is when these data

407
00:20:28,320 --> 00:20:31,360
Speaker 5: centers come in and the balance sheets of these hyperscalers

408
00:20:31,359 --> 00:20:34,280
Speaker 5: come into an area that's largely been overlooked.

409
00:20:34,359 --> 00:20:35,320
Speaker 4: You know, it's that.

410
00:20:35,320 --> 00:20:39,440
Speaker 5: New Yorker magazine cover where you know you're looking from.

411
00:20:39,600 --> 00:20:43,280
Speaker 2: The Manhattan and then it's California, right Exactly.

412
00:20:43,400 --> 00:20:47,600
Speaker 5: A lot of these areas are underinvested, particularly in infrastructure,

413
00:20:48,000 --> 00:20:50,840
Speaker 5: and a hyperscaler comes in and starts saying, well, you know,

414
00:20:51,119 --> 00:20:54,399
Speaker 5: this water treatment facility isn't sufficient, we need to build

415
00:20:54,400 --> 00:20:58,040
Speaker 5: it out. That's very meaningful part of that interconnection study

416
00:20:58,040 --> 00:21:00,399
Speaker 5: that we're doing. You're looking at the electric trans mission

417
00:21:00,440 --> 00:21:04,560
Speaker 5: of the area. That is very meaningful. But water for

418
00:21:04,640 --> 00:21:08,480
Speaker 5: the developer, it's not a problem unless it's actually a problem.

419
00:21:08,600 --> 00:21:10,879
Speaker 5: What I mean by that is we have a lawyer

420
00:21:10,920 --> 00:21:14,520
Speaker 5: on several lawyers on staff that focus on water, and

421
00:21:14,880 --> 00:21:17,440
Speaker 5: one of them, for example, knows all ninety eight water

422
00:21:17,480 --> 00:21:20,680
Speaker 5: districts in Texas and knows which one you are supposed

423
00:21:20,720 --> 00:21:22,719
Speaker 5: to go to if you want to get your project

424
00:21:22,720 --> 00:21:27,160
Speaker 5: done quickly and which one to avoid. That's something that

425
00:21:27,359 --> 00:21:30,400
Speaker 5: if you mess it up, could really affect your timeline.

426
00:21:30,480 --> 00:21:33,399
Speaker 3: Wait now, I'm really curious what would make one water

427
00:21:33,480 --> 00:21:36,199
Speaker 3: destination more attractive versus the one you want to avoid.

428
00:21:36,280 --> 00:21:39,000
Speaker 3: Is it just regulatory hurdles or is it like quality

429
00:21:39,080 --> 00:21:39,639
Speaker 3: of the water.

430
00:21:40,040 --> 00:21:42,679
Speaker 5: So it's not quality of the water. You know, if

431
00:21:42,800 --> 00:21:46,560
Speaker 5: it's pottable water, it's pottable water. It's both water supply

432
00:21:46,880 --> 00:21:49,760
Speaker 5: and then water off take. So one of the issues

433
00:21:49,760 --> 00:21:51,719
Speaker 5: in oil and gas and Texas that they've had is

434
00:21:51,920 --> 00:21:53,000
Speaker 5: where do we get all the water?

435
00:21:53,119 --> 00:21:55,159
Speaker 4: And what are we doing with all of this wastewater.

436
00:21:55,400 --> 00:21:58,400
Speaker 5: It's the exact same thing with data centers. Obviously they're

437
00:21:58,400 --> 00:22:01,000
Speaker 5: putting different chemicals into the water in order to put

438
00:22:01,000 --> 00:22:04,159
Speaker 5: it into the facility, but it's still a chemically treated

439
00:22:04,200 --> 00:22:06,720
Speaker 5: water that has to be processed before it can go

440
00:22:06,800 --> 00:22:11,119
Speaker 5: back either into the drinking supply or into the rivers

441
00:22:11,119 --> 00:22:11,680
Speaker 5: and streams.

442
00:22:11,960 --> 00:22:15,359
Speaker 2: And so, just back to the political environment, like what's

443
00:22:15,400 --> 00:22:18,080
Speaker 2: that doing on the ground today, This big change that's

444
00:22:18,080 --> 00:22:20,760
Speaker 2: occurred because everyone's up in arms about all this stuff,

445
00:22:20,840 --> 00:22:23,720
Speaker 2: how is that affecting some of the project planning that

446
00:22:23,800 --> 00:22:24,479
Speaker 2: exists today.

447
00:22:24,800 --> 00:22:27,920
Speaker 5: So I think that the developers have been very mindful

448
00:22:28,000 --> 00:22:30,320
Speaker 5: in the past, but they're even more so now because

449
00:22:30,320 --> 00:22:33,800
Speaker 5: of the sensitivity around it. In certain states, they are

450
00:22:33,840 --> 00:22:38,160
Speaker 5: considering proposals that might involve moratoriums on data centers. That's

451
00:22:38,200 --> 00:22:40,320
Speaker 5: not the case in taxas. If you look at what

452
00:22:40,560 --> 00:22:43,399
Speaker 5: Governor Abbott and the legislature have been doing, they're pushing

453
00:22:43,440 --> 00:22:48,240
Speaker 5: a more data centers, more tax incentives, more infrastructure. I

454
00:22:48,240 --> 00:22:51,240
Speaker 5: think Virginia and several other places realize how important it

455
00:22:51,280 --> 00:22:53,760
Speaker 5: is to their economies as well and to their tax

456
00:22:53,840 --> 00:22:57,560
Speaker 5: base because data centers produce a lot of tax revenue

457
00:22:57,600 --> 00:22:59,480
Speaker 5: for these states, which go to help fund.

458
00:22:59,320 --> 00:23:00,000
Speaker 4: Schools in the laws.

459
00:23:00,560 --> 00:23:04,160
Speaker 3: Is anyone talking about public private partnerships in the data

460
00:23:04,200 --> 00:23:06,600
Speaker 3: center context. I feel like this is something that comes

461
00:23:06,640 --> 00:23:10,159
Speaker 3: in waves. People start getting really really excited about, like

462
00:23:10,440 --> 00:23:13,720
Speaker 3: public private partnerships for infrastructure build out, and then you

463
00:23:13,760 --> 00:23:15,480
Speaker 3: don't hear about it for like five years, and then

464
00:23:15,520 --> 00:23:17,840
Speaker 3: it comes back. Are we in one of those waves

465
00:23:17,920 --> 00:23:20,000
Speaker 3: right now when it comes to data center spend or not?

466
00:23:20,080 --> 00:23:23,520
Speaker 5: Really? Yes, So I'm thinking about what I can say.

467
00:23:23,880 --> 00:23:26,919
Speaker 5: So the short answer is yes, those are being explored.

468
00:23:27,320 --> 00:23:30,359
Speaker 5: They're both from the perspective of public private partnerships, but

469
00:23:30,480 --> 00:23:33,520
Speaker 5: also from the perspective of like the Department of Energy

470
00:23:33,560 --> 00:23:36,280
Speaker 5: Loan Program Office and right you worked on TARA are

471
00:23:36,280 --> 00:23:39,879
Speaker 5: the others. Yes, they can provide financing support and the

472
00:23:39,920 --> 00:23:43,480
Speaker 5: Department of Energy has been very vocal about what the

473
00:23:43,560 --> 00:23:48,440
Speaker 5: Loan Program Office offers. The Department of Defense, Department of War,

474
00:23:48,480 --> 00:23:51,760
Speaker 5: I guess now also has programs that they can make available.

475
00:23:52,160 --> 00:23:55,400
Speaker 5: Some of these are actual grants, some of them are loans,

476
00:23:55,800 --> 00:23:59,479
Speaker 5: but separated apart from that, there's loan guarantees and so

477
00:23:59,640 --> 00:24:02,639
Speaker 5: you have a backstop by the federal government of a

478
00:24:02,720 --> 00:24:06,199
Speaker 5: loan that Let's be honest, if you're at forty percent

479
00:24:06,280 --> 00:24:10,119
Speaker 5: loan to value and you've got investment grade tenant that's

480
00:24:10,520 --> 00:24:12,800
Speaker 5: going to be paying the lease longer than the term

481
00:24:13,080 --> 00:24:15,439
Speaker 5: of the bond, do you really need it? Probably not,

482
00:24:15,600 --> 00:24:18,520
Speaker 5: but it can help lower the cost of capital anyway.

483
00:24:18,240 --> 00:24:21,760
Speaker 2: Actually say more about that because we did several episodes.

484
00:24:22,040 --> 00:24:24,960
Speaker 2: We've done several episodes about the idea of loan guarantees

485
00:24:25,000 --> 00:24:27,639
Speaker 2: or particularly the Loan Program's Office. Well, but from the

486
00:24:27,720 --> 00:24:30,680
Speaker 2: private side, like talk to us about how that functionally

487
00:24:30,800 --> 00:24:33,120
Speaker 2: turns into Okay, we're going to get a better credit rating,

488
00:24:33,240 --> 00:24:35,840
Speaker 2: or this is going to crowd in private capital. As

489
00:24:35,880 --> 00:24:38,320
Speaker 2: they like to say, what happens, what is the steps

490
00:24:38,400 --> 00:24:41,840
Speaker 2: via which the government's role suddenly unlocks this finance.

491
00:24:42,240 --> 00:24:45,359
Speaker 5: I think in all credit, when you're trying to underwrite

492
00:24:45,640 --> 00:24:47,800
Speaker 5: any kind of loan, you want to see what are

493
00:24:47,840 --> 00:24:50,600
Speaker 5: the cash flows or what's the collateral? And am I

494
00:24:50,640 --> 00:24:53,119
Speaker 5: going to get paid back? What's the likeliod right? And

495
00:24:53,200 --> 00:24:56,720
Speaker 5: one of the things that the frameworks I think are

496
00:24:57,119 --> 00:24:58,720
Speaker 5: show me who you walk with, and I'll show you

497
00:24:58,760 --> 00:24:59,200
Speaker 5: who you are.

498
00:24:59,359 --> 00:24:59,600
Speaker 4: Right.

499
00:25:00,240 --> 00:25:03,439
Speaker 5: Fine, if you have the United States government that is

500
00:25:03,480 --> 00:25:05,400
Speaker 5: providing a backstop, then I can be.

501
00:25:05,359 --> 00:25:06,960
Speaker 4: Confident I'm going to get paid back.

502
00:25:07,320 --> 00:25:11,919
Speaker 5: If it's Microsoft or Google or another who's providing a

503
00:25:12,160 --> 00:25:16,000
Speaker 5: guarantee on a project or the loan, then I'm very

504
00:25:16,040 --> 00:25:18,399
Speaker 5: comfortable that I'm going to get paid back. When you

505
00:25:18,440 --> 00:25:21,919
Speaker 5: put these types of things together, that means I don't

506
00:25:21,960 --> 00:25:26,280
Speaker 5: have to charge you an extra incremental amount to protect

507
00:25:26,280 --> 00:25:29,439
Speaker 5: me from that credit risk, and that extra money that

508
00:25:29,480 --> 00:25:32,200
Speaker 5: you keep in your pocket you can then turn into

509
00:25:32,480 --> 00:25:36,680
Speaker 5: additional projects and it's a velocity of money concept.

510
00:25:37,320 --> 00:25:39,840
Speaker 3: I want to go back to private credit for a second, because,

511
00:25:40,080 --> 00:25:42,159
Speaker 3: as you mentioned earlier, a lot of these deals have

512
00:25:42,240 --> 00:25:45,320
Speaker 3: been done in the private credit space so far, and

513
00:25:45,960 --> 00:25:48,600
Speaker 3: companies have a lot of options when it comes to financing.

514
00:25:48,640 --> 00:25:50,320
Speaker 3: As I mentioned in the intro, they can go the

515
00:25:50,320 --> 00:25:54,080
Speaker 3: public route, they can securitize, they can go to private credit.

516
00:25:54,520 --> 00:25:56,880
Speaker 3: And whenever people talk about private credit in the context

517
00:25:56,880 --> 00:26:00,480
Speaker 3: of AI, they always say vague things like, oh, provides

518
00:26:00,600 --> 00:26:05,560
Speaker 3: customizable financing options and stuff like that. What exactly is

519
00:26:05,600 --> 00:26:09,399
Speaker 3: the attraction of private credit for data centers and AI

520
00:26:09,840 --> 00:26:10,360
Speaker 3: build out.

521
00:26:10,960 --> 00:26:14,439
Speaker 5: So one of the most attractive things that private credit

522
00:26:14,520 --> 00:26:18,360
Speaker 5: can but doesn't always offer, is nondluted capital. Right, equity

523
00:26:18,440 --> 00:26:22,680
Speaker 5: is diluted, and equity is very expensive. There's that risk premium.

524
00:26:22,280 --> 00:26:23,439
Speaker 4: That's associated with it.

525
00:26:23,840 --> 00:26:27,400
Speaker 5: Private credit, there is an assumption in the story that

526
00:26:28,080 --> 00:26:31,680
Speaker 5: because of the extra protections that they receive, they don't

527
00:26:31,680 --> 00:26:34,760
Speaker 5: need the same equity risk premium. So a lot of

528
00:26:34,840 --> 00:26:37,600
Speaker 5: them are looking at maybe a fifteen percent IRR. They

529
00:26:37,680 --> 00:26:40,919
Speaker 5: might charge you ten to twelve percent up front, but

530
00:26:40,960 --> 00:26:45,080
Speaker 5: then because of the moint that basically their minimum return

531
00:26:45,160 --> 00:26:47,880
Speaker 5: on the capital that they'll get paid or an IRR

532
00:26:48,000 --> 00:26:51,639
Speaker 5: premium over time in the takeout, then they'll have a

533
00:26:51,760 --> 00:26:55,200
Speaker 5: catch up on the back end the equity they're taking

534
00:26:55,200 --> 00:26:57,280
Speaker 5: a percentage of what you at the end of the

535
00:26:57,359 --> 00:26:59,040
Speaker 5: day are going to get, and it's usually a much

536
00:26:59,080 --> 00:27:01,080
Speaker 5: more expensive piece private credit.

537
00:27:01,359 --> 00:27:04,080
Speaker 2: But what about private credit versus other forms of credit

538
00:27:04,280 --> 00:27:06,160
Speaker 2: like bank lending or the public bond market.

539
00:27:06,359 --> 00:27:07,280
Speaker 4: Well, can you get it?

540
00:27:07,560 --> 00:27:09,600
Speaker 2: Well, I mean that's I mean, what is that's a

541
00:27:09,960 --> 00:27:10,359
Speaker 2: I don't know.

542
00:27:10,560 --> 00:27:14,640
Speaker 5: The underwriting standards for private credit can be riskier than

543
00:27:14,680 --> 00:27:17,360
Speaker 5: what you would have from you know, a bull bracket

544
00:27:17,400 --> 00:27:18,800
Speaker 5: bank's credit desk.

545
00:27:34,760 --> 00:27:38,160
Speaker 3: So when it comes to tenant diversification in data centers,

546
00:27:38,640 --> 00:27:42,959
Speaker 3: how is that judged and evaluated and priced? Because if

547
00:27:43,000 --> 00:27:45,439
Speaker 3: I think about who's you know, using a data center,

548
00:27:46,160 --> 00:27:49,320
Speaker 3: correlation is hard to measure at the best of times.

549
00:27:49,359 --> 00:27:50,960
Speaker 3: And if I think about, you know, like a new

550
00:27:51,000 --> 00:27:54,399
Speaker 3: business that's growing and everyone is suddenly using it, it

551
00:27:54,560 --> 00:27:56,719
Speaker 3: feels like the kind of thing that could end up

552
00:27:56,720 --> 00:27:59,200
Speaker 3: being very correlated as opposed to diversified.

553
00:27:59,800 --> 00:28:05,600
Speaker 5: So when you are thinking about larger investment grade bonds

554
00:28:05,720 --> 00:28:09,639
Speaker 5: and debt in data centers, normally you are thinking about

555
00:28:09,920 --> 00:28:14,919
Speaker 5: wholesale data centers where there's one tenant that has that facility,

556
00:28:15,359 --> 00:28:20,040
Speaker 5: and that tenant has investment grade credit qual it's much

557
00:28:20,119 --> 00:28:22,719
Speaker 5: less likely that you are going to have a pool

558
00:28:22,800 --> 00:28:26,199
Speaker 5: of non investment grade tenants in a kind of co

559
00:28:26,320 --> 00:28:31,520
Speaker 5: location style data center, and then because of that, Fitch

560
00:28:31,600 --> 00:28:34,160
Speaker 5: or S and P or Kroll is going to give

561
00:28:34,200 --> 00:28:37,439
Speaker 5: you this investment grade rating. So I don't think that

562
00:28:37,520 --> 00:28:41,880
Speaker 5: the expectation should be that you're going to get a

563
00:28:42,320 --> 00:28:46,000
Speaker 5: investment grade rating on a pool of non investment grade tenants,

564
00:28:46,080 --> 00:28:48,600
Speaker 5: at least not right now. I think that eventually, when

565
00:28:48,640 --> 00:28:52,960
Speaker 5: you've got thousands of tenants across hundreds of data centers,

566
00:28:53,240 --> 00:28:54,760
Speaker 5: that becomes much more attractive.

567
00:28:54,880 --> 00:28:57,080
Speaker 4: It's just not really where the market is right now.

568
00:28:57,160 --> 00:28:59,520
Speaker 2: We did an episode a few months ago in September

569
00:28:59,520 --> 00:29:02,160
Speaker 2: actually where we're talking to this guy, Don Wilson in

570
00:29:02,240 --> 00:29:06,160
Speaker 2: Chicago who setting up a futures trading firm for GPUs

571
00:29:06,240 --> 00:29:09,000
Speaker 2: and theoretically it can hedge your GPU costs. As you

572
00:29:09,040 --> 00:29:11,880
Speaker 2: think about the industry going forward, could you see that

573
00:29:12,000 --> 00:29:16,760
Speaker 2: being a useful instrument for data center financing. Where I'm

574
00:29:16,760 --> 00:29:20,560
Speaker 2: a little concerned about is the value of this collateral goodness, name,

575
00:29:20,640 --> 00:29:23,640
Speaker 2: I want to hedge the GPU exposure. Could you see

576
00:29:23,640 --> 00:29:26,320
Speaker 2: over time maturing and debt being a valuable thing.

577
00:29:26,760 --> 00:29:29,800
Speaker 5: I love new finance tools. I think that that kind

578
00:29:29,800 --> 00:29:34,040
Speaker 5: of creativity is really important. There's always an opportunity there

579
00:29:34,520 --> 00:29:38,040
Speaker 5: that The thing I have noticed with the data centers,

580
00:29:38,120 --> 00:29:41,280
Speaker 5: just with other asset classes, is that there's always a

581
00:29:41,320 --> 00:29:45,560
Speaker 5: cash flow stream that is not being utilized. And so

582
00:29:45,680 --> 00:29:49,280
Speaker 5: for example, if you have a securitization you're or really

583
00:29:49,320 --> 00:29:52,400
Speaker 5: any bond, you're not necessarily going to get credit for

584
00:29:52,720 --> 00:29:56,960
Speaker 5: everything in it. So your contracted cash flows, if that

585
00:29:57,040 --> 00:30:00,479
Speaker 5: counter party is an investment grade, may not get an

586
00:30:00,520 --> 00:30:04,320
Speaker 5: advance rate a loan to value on that. Somebody else

587
00:30:04,360 --> 00:30:06,960
Speaker 5: should be there in order to take advantage of that

588
00:30:07,160 --> 00:30:09,880
Speaker 5: and provide you with additional capital that you can then

589
00:30:10,000 --> 00:30:11,360
Speaker 5: reinvest into development.

590
00:30:12,360 --> 00:30:14,880
Speaker 3: So last week we had a big outage at the

591
00:30:14,880 --> 00:30:18,560
Speaker 3: CME and futures were basically frozen for like ten hours

592
00:30:18,680 --> 00:30:20,920
Speaker 3: or something, and this was because of an issue at

593
00:30:20,960 --> 00:30:24,040
Speaker 3: one of the CME's data centers, which is operated by

594
00:30:24,120 --> 00:30:28,560
Speaker 3: a company called cyrus One. Is there a reputational or

595
00:30:28,640 --> 00:30:32,840
Speaker 3: operational risk that either deal structures or investors need to

596
00:30:32,840 --> 00:30:35,880
Speaker 3: be aware of when it comes to these financing arrangements.

597
00:30:35,680 --> 00:30:38,360
Speaker 5: Reputational or in what regards so well.

598
00:30:38,400 --> 00:30:41,520
Speaker 3: For instance, if cyrus one has a big melt down

599
00:30:41,640 --> 00:30:43,719
Speaker 3: at one of their data centers which seems to have

600
00:30:43,760 --> 00:30:48,440
Speaker 3: happened because one of their cooling centers reportedly malfunctioned. Is

601
00:30:48,440 --> 00:30:52,720
Speaker 3: that something that then gets priced into the financing arrangement

602
00:30:52,880 --> 00:30:56,120
Speaker 3: or is that something that investors should be concerned about.

603
00:30:56,480 --> 00:31:00,680
Speaker 5: So these are definitely things that investors underwright too. So

604
00:31:01,160 --> 00:31:06,320
Speaker 5: if you are a blue chip operator and developer, you

605
00:31:06,400 --> 00:31:09,040
Speaker 5: are going to have a lower cost of capital than

606
00:31:09,080 --> 00:31:12,160
Speaker 5: you would otherwise. There are a lot of Johnny cum

607
00:31:12,240 --> 00:31:15,720
Speaker 5: Lateli's into data centers over the last two years, since

608
00:31:15,760 --> 00:31:19,360
Speaker 5: AI has really come to the forefront. The people that

609
00:31:19,400 --> 00:31:21,320
Speaker 5: have been doing it for a decade, they know what

610
00:31:21,360 --> 00:31:24,160
Speaker 5: they're doing and they're very good at it. And that

611
00:31:24,200 --> 00:31:27,560
Speaker 5: doesn't mean that there aren't going to be problems. Things

612
00:31:27,600 --> 00:31:29,400
Speaker 5: blow up sometimes.

613
00:31:28,800 --> 00:31:32,320
Speaker 3: Squirrels, two wires, yeah, hopefully not subsea ones.

614
00:31:33,400 --> 00:31:36,760
Speaker 5: There are accidents all the time, but that's part of

615
00:31:36,800 --> 00:31:41,600
Speaker 5: why you have these pretty incredible engineering studies that are

616
00:31:41,640 --> 00:31:44,160
Speaker 5: done in order to actually put these things together, and

617
00:31:44,240 --> 00:31:48,560
Speaker 5: freak accidents happen, and you know, that's part of the

618
00:31:48,600 --> 00:31:49,200
Speaker 5: credit risk.

619
00:31:49,480 --> 00:31:53,720
Speaker 2: We just compared to other sort of real estate plays,

620
00:31:54,200 --> 00:31:56,840
Speaker 2: is there more sort of I guess, operational risk in

621
00:31:56,920 --> 00:31:59,760
Speaker 2: a data center than in other times? You know, we're

622
00:31:59,840 --> 00:32:04,280
Speaker 2: just some sort of retail distribution facility. I imagine the

623
00:32:04,320 --> 00:32:06,840
Speaker 2: sort of operational risk isn't going to be as great

624
00:32:06,880 --> 00:32:07,680
Speaker 2: and something like that.

625
00:32:07,840 --> 00:32:11,000
Speaker 5: Yeah, so that's part of why they have these service

626
00:32:11,080 --> 00:32:14,880
Speaker 5: level agreements within the data centers themselves, where the operator

627
00:32:15,120 --> 00:32:20,480
Speaker 5: is agreeing to provide an uninterruptible, non intermittent power. That's

628
00:32:20,520 --> 00:32:23,280
Speaker 5: one of the big things that people focus on. And

629
00:32:23,440 --> 00:32:26,080
Speaker 5: how do the terms of that contract work and what

630
00:32:26,120 --> 00:32:30,040
Speaker 5: are the backstops. You could have a reserve account associated

631
00:32:30,040 --> 00:32:32,200
Speaker 5: with it so that if there are payments that are

632
00:32:32,240 --> 00:32:35,400
Speaker 5: necessary to be made, that there's just cash sitting there

633
00:32:35,840 --> 00:32:39,000
Speaker 5: ready to go. Normally, people are thinking that that would

634
00:32:39,080 --> 00:32:42,040
Speaker 5: reduce the cash flow on the bond, and that would

635
00:32:42,080 --> 00:32:46,120
Speaker 5: reduce at the end of the day, the equity distribution coming.

636
00:32:45,840 --> 00:32:50,080
Speaker 4: From the SPV as opposed to affecting the payment on

637
00:32:50,120 --> 00:32:50,480
Speaker 4: the bond.

638
00:32:50,800 --> 00:32:53,719
Speaker 3: That reminds me, actually, are you seeing a big surge

639
00:32:53,720 --> 00:32:57,600
Speaker 3: and demand for data center insurance or for people you

640
00:32:57,600 --> 00:33:00,000
Speaker 3: know who want to ensure not just against operational risks,

641
00:33:00,200 --> 00:33:02,200
Speaker 3: but cyber risk and things like that.

642
00:33:02,560 --> 00:33:02,760
Speaker 4: Yes.

643
00:33:03,000 --> 00:33:05,920
Speaker 5: So one of the wonderful things about any kind of

644
00:33:05,960 --> 00:33:09,080
Speaker 5: economic activity is that the insurance market is always there.

645
00:33:08,920 --> 00:33:09,560
Speaker 4: To support it.

646
00:33:10,280 --> 00:33:11,280
Speaker 3: Like the lawyers as well.

647
00:33:11,360 --> 00:33:13,560
Speaker 5: Right, yeah, definitely give us a call.

648
00:33:13,640 --> 00:33:13,880
Speaker 4: Right.

649
00:33:14,000 --> 00:33:17,480
Speaker 5: The reality though, is there are products that data centers

650
00:33:17,480 --> 00:33:20,920
Speaker 5: have had for decades, and then there are new risks

651
00:33:20,960 --> 00:33:23,720
Speaker 5: that are coming up just because of how AI training works.

652
00:33:24,120 --> 00:33:25,960
Speaker 5: You can always go to Lloyd's of London in order

653
00:33:26,000 --> 00:33:28,920
Speaker 5: to get a specific policy if necessary, But for the

654
00:33:28,920 --> 00:33:31,560
Speaker 5: most part, these are risks that have already been priced in,

655
00:33:31,600 --> 00:33:34,600
Speaker 5: and there are products that are already there. Now there's

656
00:33:34,720 --> 00:33:38,280
Speaker 5: talk of well, there are products that exist in terms

657
00:33:38,280 --> 00:33:42,320
Speaker 5: of getting an insurance product to handle your technology risk.

658
00:33:42,680 --> 00:33:44,680
Speaker 5: So does the product work at the end of the

659
00:33:44,760 --> 00:33:48,480
Speaker 5: day or does it become obsolete too quickly? That looks

660
00:33:48,560 --> 00:33:52,200
Speaker 5: more like a bet, almost as opposed to a real

661
00:33:52,240 --> 00:33:53,120
Speaker 5: insurance policy.

662
00:33:53,520 --> 00:33:55,400
Speaker 4: But it's something that can be underwritten.

663
00:33:55,920 --> 00:33:58,400
Speaker 2: How valuable it is just having a plug into the grid,

664
00:33:58,480 --> 00:34:01,200
Speaker 2: you know, I'm thinking about are those various deals of

665
00:34:01,560 --> 00:34:05,920
Speaker 2: companies wanting to buy bitcoin mining operations, for example, and

666
00:34:06,000 --> 00:34:08,480
Speaker 2: it seems like you know what, yeah, maybe you can

667
00:34:08,520 --> 00:34:10,680
Speaker 2: make a little money mining bitcoin, but you have you

668
00:34:10,760 --> 00:34:13,799
Speaker 2: have access to twenty four to seven reliable power. That's

669
00:34:13,840 --> 00:34:17,000
Speaker 2: a lot more valuable than what's going on inside the

670
00:34:17,000 --> 00:34:19,680
Speaker 2: shell here. Talk to us just about that value of

671
00:34:19,719 --> 00:34:21,400
Speaker 2: anyone who has access to power.

672
00:34:21,480 --> 00:34:27,239
Speaker 5: So powered land is huge and that is developing. You know,

673
00:34:27,719 --> 00:34:30,319
Speaker 5: it's existed for a while, but is becoming more and

674
00:34:30,360 --> 00:34:33,960
Speaker 5: more important because there are rets and real estate developers

675
00:34:34,280 --> 00:34:37,319
Speaker 5: who don't need to have the sophistication of can I

676
00:34:37,400 --> 00:34:41,600
Speaker 5: do the new direct to cooling technology for a powered

677
00:34:41,600 --> 00:34:44,880
Speaker 5: shell or turnkey solution. I'm just going to have land

678
00:34:44,960 --> 00:34:46,800
Speaker 5: and I'm going to make sure that there's sufficient power

679
00:34:46,800 --> 00:34:49,799
Speaker 5: here for you to build your data center or for

680
00:34:49,880 --> 00:34:54,120
Speaker 5: you to bring your GPUs into. That's attractive if you

681
00:34:54,200 --> 00:34:57,040
Speaker 5: listen to the earnings calls of several of the real

682
00:34:57,120 --> 00:35:00,440
Speaker 5: estate developers, they're moving into that if they have an already.

683
00:35:00,880 --> 00:35:03,600
Speaker 2: So their job is they're just going to secure land

684
00:35:03,600 --> 00:35:05,839
Speaker 2: that they can that they know they can get power too.

685
00:35:05,880 --> 00:35:07,640
Speaker 5: So they've already got like they've got that right. So

686
00:35:08,000 --> 00:35:11,480
Speaker 5: the interesting part is there's so many different projects over

687
00:35:11,520 --> 00:35:15,440
Speaker 5: the last several decades where they've been working through interconnection

688
00:35:15,560 --> 00:35:19,239
Speaker 5: studies for the land for you name it project, and

689
00:35:19,600 --> 00:35:22,719
Speaker 5: they're waiting years and years just like everybody else, in

690
00:35:22,800 --> 00:35:25,759
Speaker 5: order to get those done. Well, why use it for

691
00:35:25,800 --> 00:35:26,879
Speaker 5: this when we can use it.

692
00:35:26,840 --> 00:35:28,560
Speaker 4: For that and data centers?

693
00:35:28,600 --> 00:35:30,760
Speaker 5: You can make a lot more money on it.

694
00:35:30,800 --> 00:35:34,439
Speaker 3: Isn't the obvious solution to the power problem just for

695
00:35:34,680 --> 00:35:37,839
Speaker 3: the big guys to build their own power system and

696
00:35:37,920 --> 00:35:40,439
Speaker 3: maybe spend a little bit more money plugging it into

697
00:35:40,480 --> 00:35:43,160
Speaker 3: the grid. Again, if political pushback is becoming an issue,

698
00:35:43,239 --> 00:35:46,040
Speaker 3: or if regulators are worried about this, shouldn't they just

699
00:35:46,080 --> 00:35:48,120
Speaker 3: do it on their own? Is that the straightforward thing

700
00:35:48,160 --> 00:35:48,440
Speaker 3: to do.

701
00:35:49,280 --> 00:35:52,399
Speaker 5: The hyperscalers themselves. I think that goes back to what's

702
00:35:52,440 --> 00:35:54,279
Speaker 5: the best use of their capital. They're not in the

703
00:35:54,320 --> 00:35:58,279
Speaker 5: infrastructure business specifically, though, I will say if you look

704
00:35:58,320 --> 00:36:03,239
Speaker 5: at Love, Amazon is an example. They sold books, they

705
00:36:03,239 --> 00:36:06,640
Speaker 5: had the website, they needed to expand, so they build

706
00:36:06,680 --> 00:36:09,080
Speaker 5: Amazon Web services. They need a distribution.

707
00:36:09,320 --> 00:36:09,920
Speaker 4: They've got that.

708
00:36:10,760 --> 00:36:15,520
Speaker 5: Tech companies have been becoming energy companies for years. Energy companies,

709
00:36:15,560 --> 00:36:18,319
Speaker 5: we know, had to become tech companies a decade ago

710
00:36:18,440 --> 00:36:21,600
Speaker 5: in order to just keep up. But now there's this

711
00:36:22,200 --> 00:36:26,000
Speaker 5: integration that's happening from both sides, both in terms of

712
00:36:26,040 --> 00:36:29,320
Speaker 5: what's happening in our operations but the ownership itself.

713
00:36:29,960 --> 00:36:33,000
Speaker 2: Just going back to powered land for a second, So

714
00:36:33,040 --> 00:36:35,560
Speaker 2: what you're saying is that there are these developments that

715
00:36:35,600 --> 00:36:37,760
Speaker 2: have been going on a long time, but they initially

716
00:36:37,800 --> 00:36:41,000
Speaker 2: planned for something else, But that today in twenty twenty five,

717
00:36:41,120 --> 00:36:43,400
Speaker 2: is they're getting closer to when they could be connected.

718
00:36:43,640 --> 00:36:46,640
Speaker 2: Maybe AI is you know, this connection is a lot

719
00:36:46,680 --> 00:36:48,920
Speaker 2: more valuable for an AI data center.

720
00:36:49,000 --> 00:36:50,280
Speaker 4: Absolutely, so this is interesting.

721
00:36:50,280 --> 00:36:52,000
Speaker 2: See this to me, like, the only reason I go

722
00:36:52,080 --> 00:36:54,480
Speaker 2: back to this is because one of the things that

723
00:36:54,520 --> 00:36:56,959
Speaker 2: when people talk about an AI bubble, if there is one,

724
00:36:57,280 --> 00:37:00,239
Speaker 2: is this idea of crowding out other productive use of

725
00:37:00,280 --> 00:37:02,640
Speaker 2: the economy. Right, are there better things in the long

726
00:37:02,719 --> 00:37:04,840
Speaker 2: term that we could have done with these turbines? Are

727
00:37:04,880 --> 00:37:07,200
Speaker 2: there better things that we could have done with these

728
00:37:07,719 --> 00:37:11,520
Speaker 2: electrical connection systems, et cetera. And other companies might have

729
00:37:11,600 --> 00:37:13,680
Speaker 2: been waiting on some piece of gear and the AI

730
00:37:13,760 --> 00:37:15,840
Speaker 2: data center outbid them, And I'm going to I'm not

731
00:37:15,880 --> 00:37:19,080
Speaker 2: asking you like this sustainable, but there are other things

732
00:37:19,239 --> 00:37:21,760
Speaker 2: that are going to lose out or not have access

733
00:37:21,800 --> 00:37:24,839
Speaker 2: to electricity that people wanted to do but aren't going

734
00:37:24,880 --> 00:37:28,440
Speaker 2: to because that wire is more valuable for an AI.

735
00:37:28,239 --> 00:37:29,440
Speaker 4: Company one hundred percent.

736
00:37:29,760 --> 00:37:33,399
Speaker 5: Think about the actual interconnection queue two and a half

737
00:37:33,440 --> 00:37:37,120
Speaker 5: years ago, before AI became a big deal, there was

738
00:37:37,840 --> 00:37:42,000
Speaker 5: I want to say, twenty six hundred gigawatts in the

739
00:37:42,040 --> 00:37:45,520
Speaker 5: interconnection queue, and we were expecting eighty percent of that

740
00:37:45,560 --> 00:37:49,120
Speaker 5: would never actually be constructed. Only twenty percent of it would.

741
00:37:49,280 --> 00:37:52,880
Speaker 5: And that's pretty typical even on a going forward basis,

742
00:37:53,560 --> 00:37:56,000
Speaker 5: what are you going to do with those projects and

743
00:37:56,520 --> 00:37:59,319
Speaker 5: that land that has already started in a process of

744
00:37:59,320 --> 00:38:01,839
Speaker 5: interconnections that he is and the like, Well, you can

745
00:38:02,160 --> 00:38:06,160
Speaker 5: shift that to data centers, especially because a lot of

746
00:38:06,160 --> 00:38:10,360
Speaker 5: that you were producing solar batteries, wind and other power

747
00:38:10,400 --> 00:38:13,239
Speaker 5: gen Now you add on the data center layer to it.

748
00:38:13,640 --> 00:38:16,759
Speaker 5: And what may not have been economic before now is.

749
00:38:17,239 --> 00:38:18,399
Speaker 3: Should I build a data center?

750
00:38:18,480 --> 00:38:19,520
Speaker 2: Joe, Yeah, I've.

751
00:38:19,360 --> 00:38:22,480
Speaker 3: Got a grid connection, I've got a water Yeah, there

752
00:38:22,480 --> 00:38:25,239
Speaker 3: we go. Okay, next project all lots builds a data

753
00:38:25,239 --> 00:38:28,480
Speaker 3: center in protected land in Connecticut?

754
00:38:29,400 --> 00:38:33,920
Speaker 6: Probably not actually, So this sort of piggybacks on a

755
00:38:34,200 --> 00:38:38,279
Speaker 6: question that Tracy as already. But why not just for

756
00:38:38,480 --> 00:38:41,600
Speaker 6: you know, these is companies, particularly the hyperscalers that look

757
00:38:41,600 --> 00:38:44,200
Speaker 6: out in the environment and there's all these people showing

758
00:38:44,239 --> 00:38:47,319
Speaker 6: up at meetings complaining about the water et cetera like that,

759
00:38:47,840 --> 00:38:52,000
Speaker 6: Why isn't the future just entirely behind the meter in

760
00:38:52,040 --> 00:38:53,920
Speaker 6: Texas where it's like we're just gonna.

761
00:38:53,680 --> 00:38:54,120
Speaker 1: Build it all.

762
00:38:54,160 --> 00:38:56,799
Speaker 2: We're gonna have the natural gas plant on site. We're

763
00:38:56,840 --> 00:38:59,160
Speaker 2: never gonna bother with the grid. We're never gonna We're

764
00:38:59,200 --> 00:39:01,279
Speaker 2: just gonna have the plan right there. Why isn't that

765
00:39:01,440 --> 00:39:02,960
Speaker 2: just the entire future of data center?

766
00:39:03,160 --> 00:39:05,440
Speaker 5: One of the issues with behind the meter is what

767
00:39:05,719 --> 00:39:09,560
Speaker 5: if the data center goes away? So, for example, what

768
00:39:09,640 --> 00:39:13,400
Speaker 5: if Mark Zuckerberg one day decides I'm not doing the

769
00:39:13,440 --> 00:39:17,640
Speaker 5: metaverse anymore and somebody in AI says I'm not doing AI,

770
00:39:18,480 --> 00:39:22,520
Speaker 5: or we move from GPUs to quantum.

771
00:39:22,080 --> 00:39:23,600
Speaker 4: Computing or something like that.

772
00:39:24,000 --> 00:39:28,200
Speaker 5: You want to have that generative capacity interconnected with the

773
00:39:28,239 --> 00:39:30,520
Speaker 5: grid so that you don't have a stranded asset.

774
00:39:30,800 --> 00:39:33,839
Speaker 3: Okay, oh yeah, stranded assets. There's a word I haven't

775
00:39:33,840 --> 00:39:34,200
Speaker 3: heard of you.

776
00:39:34,280 --> 00:39:34,640
Speaker 1: Yeah, I know.

777
00:39:35,160 --> 00:39:39,040
Speaker 2: So just looking forward and I mentioned earlier core Weave

778
00:39:39,200 --> 00:39:41,319
Speaker 2: saying okay, there is a delay and some of it's

779
00:39:41,360 --> 00:39:43,880
Speaker 2: build out. What are the big choke points and do

780
00:39:43,880 --> 00:39:46,399
Speaker 2: you expect them to stack up? We recently did an

781
00:39:46,440 --> 00:39:50,160
Speaker 2: episode with a Travis Cavula Energy and he was like,

782
00:39:50,239 --> 00:39:52,720
Speaker 2: I don't know, like the amount of just the sheer

783
00:39:52,760 --> 00:39:56,000
Speaker 2: amount that we're adding to the grid, Like it's gonna

784
00:39:56,040 --> 00:39:58,000
Speaker 2: be tough and I don't know how it's all going

785
00:39:58,080 --> 00:40:00,279
Speaker 2: to pan out for some of these projects, but what

786
00:40:00,320 --> 00:40:02,479
Speaker 2: do you see as the sort of big bottle extra

787
00:40:02,600 --> 00:40:04,359
Speaker 2: choke points that you're thinking about in the coming years.

788
00:40:04,840 --> 00:40:07,840
Speaker 5: So power continues to be the number one bottle, like

789
00:40:08,080 --> 00:40:11,560
Speaker 5: I don't think anybody would dispute that. Water obviously is

790
00:40:11,560 --> 00:40:15,080
Speaker 5: an issue as well. Getting the turbines and getting what

791
00:40:15,120 --> 00:40:19,080
Speaker 5: you actually need in order to produce power is very difficult.

792
00:40:19,680 --> 00:40:22,040
Speaker 5: I do think though, that when you look at how

793
00:40:22,120 --> 00:40:27,360
Speaker 5: some of these interconnection requests are prepared, you have five

794
00:40:27,680 --> 00:40:32,240
Speaker 5: or ten different people making applications for the exact same project,

795
00:40:32,719 --> 00:40:36,440
Speaker 5: So it inflates what the expectations are on the number

796
00:40:36,480 --> 00:40:39,200
Speaker 5: of projects in the market at a given time. When

797
00:40:39,280 --> 00:40:42,480
Speaker 5: you get rid of all of that extra wash, you

798
00:40:42,520 --> 00:40:49,360
Speaker 5: wind up with hyperscalers, large enterprises, and other real investment

799
00:40:49,400 --> 00:40:53,120
Speaker 5: grade or serious tenets. Give you projects that can get done.

800
00:40:53,239 --> 00:40:56,600
Speaker 5: That's infrastructure, that's infrastructure grade, and a lot of the

801
00:40:56,719 --> 00:41:01,160
Speaker 5: speculative assets those are power points. They probably are not

802
00:41:01,200 --> 00:41:03,440
Speaker 5: going to get made, particularly if there's any kind of

803
00:41:03,560 --> 00:41:04,640
Speaker 5: economic shock later.

804
00:41:05,280 --> 00:41:07,680
Speaker 2: Travis Watford, thank you so much for coming on OUTLAS.

805
00:41:07,760 --> 00:41:08,160
Speaker 2: It was great.

806
00:41:08,560 --> 00:41:09,879
Speaker 3: Thank you, Thank you so much.

807
00:41:10,040 --> 00:41:25,200
Speaker 2: Appreciate it, Tracy. I thought that was really helpful, really

808
00:41:25,239 --> 00:41:28,640
Speaker 2: clarified a lot of things for me. I'm trying to think,

809
00:41:28,800 --> 00:41:32,560
Speaker 2: Like the point about the sequencing of the financing, I

810
00:41:32,600 --> 00:41:34,920
Speaker 2: thought was really interesting or very important to help me

811
00:41:35,000 --> 00:41:37,480
Speaker 2: understand these things, because in my mind, I'm like, well,

812
00:41:37,480 --> 00:41:39,680
Speaker 2: they're going to lose so much money. They're breaking ground

813
00:41:39,680 --> 00:41:41,880
Speaker 2: on all these deals. What if they don't actually get

814
00:41:41,920 --> 00:41:44,359
Speaker 2: implemented because you're waiting forever? And so you're hearing him

815
00:41:44,400 --> 00:41:46,960
Speaker 2: describe the sequencing of different financing at different stages, like,

816
00:41:46,960 --> 00:41:49,680
Speaker 2: all right, well, at least that makes sense to me.

817
00:41:50,239 --> 00:41:54,200
Speaker 3: Yes, I still feel like there's kind of a mismatch issue.

818
00:41:54,440 --> 00:41:55,600
Speaker 2: Sure, well, it.

819
00:41:55,680 --> 00:41:58,759
Speaker 3: Just feels like, you know, you're talking about a a

820
00:41:58,800 --> 00:42:03,600
Speaker 3: technology that has like its own upgrade risk. Let's say,

821
00:42:03,640 --> 00:42:06,680
Speaker 3: like you know, people are developing new chips pretty fast,

822
00:42:06,800 --> 00:42:08,919
Speaker 3: and you don't know when the next one is coming

823
00:42:08,920 --> 00:42:10,720
Speaker 3: down the line and when you might want to replace

824
00:42:10,800 --> 00:42:13,000
Speaker 3: all your chips with something else. So that's one thing

825
00:42:13,680 --> 00:42:16,399
Speaker 3: tendency rollover risk. I know he pushed back a little

826
00:42:16,400 --> 00:42:19,400
Speaker 3: bit on the diversified point, but again, my understanding is

827
00:42:19,400 --> 00:42:22,920
Speaker 3: for a lot of the ABS structures, maybe CMBs as well,

828
00:42:23,360 --> 00:42:26,640
Speaker 3: diversification is part of the proposal, and I think he

829
00:42:26,719 --> 00:42:30,480
Speaker 3: mentioned it earlier, and that seems difficult to me to

830
00:42:30,600 --> 00:42:33,520
Speaker 3: accurately measure. If you have a bunch of tenants all

831
00:42:33,760 --> 00:42:36,719
Speaker 3: you know, doing something in the cloud, drawing something from

832
00:42:36,800 --> 00:42:40,319
Speaker 3: a data center, if there's a big macroeconomic downturn or

833
00:42:40,360 --> 00:42:42,560
Speaker 3: something like, who's to say that they're not all going

834
00:42:42,640 --> 00:42:44,440
Speaker 3: to renig on their lease at once.

835
00:42:44,640 --> 00:42:48,879
Speaker 2: The levels of uncertainty yah, seems so extreme because you're

836
00:42:48,920 --> 00:42:52,440
Speaker 2: talking about, Okay, there's the economic downturn. There's the fact

837
00:42:52,440 --> 00:42:55,200
Speaker 2: that maybe a lot of this AI stuff could completely

838
00:42:55,239 --> 00:42:59,520
Speaker 2: fizzle out and doesn't produce even in normal times. Then

839
00:42:59,560 --> 00:43:03,600
Speaker 2: there's the technological questions. Then there's the operational questions about

840
00:43:03,600 --> 00:43:06,000
Speaker 2: are you actually good at operating and building a data center?

841
00:43:06,120 --> 00:43:08,200
Speaker 2: Not everyone who's gonna be the same, And then there's

842
00:43:08,280 --> 00:43:11,360
Speaker 2: the grid interconnection and all of these things about like

843
00:43:11,480 --> 00:43:15,439
Speaker 2: reliability of power, et cetera. So it feels like, yes,

844
00:43:15,960 --> 00:43:20,600
Speaker 2: on the one hand, I'd very much buy that on paper. Yes,

845
00:43:20,880 --> 00:43:23,759
Speaker 2: you know, this is the twenty twenty five iteration of

846
00:43:23,800 --> 00:43:25,840
Speaker 2: what used to be cell towers or what used to

847
00:43:25,840 --> 00:43:29,400
Speaker 2: be rooftop solar or anything else. But with just an.

848
00:43:29,719 --> 00:43:31,720
Speaker 2: I mean he mentioned with the cell tower for example,

849
00:43:31,920 --> 00:43:34,000
Speaker 2: if one guy whose job is to mow the lawn, right,

850
00:43:35,640 --> 00:43:41,040
Speaker 2: that's like that's the sort the like operational component. I

851
00:43:41,040 --> 00:43:43,440
Speaker 2: get the you know, most of the time the tower

852
00:43:43,520 --> 00:43:45,560
Speaker 2: is just there, right, and you have to make sure

853
00:43:45,600 --> 00:43:49,600
Speaker 2: that it's in it's safe area, et cetera. But the

854
00:43:49,760 --> 00:43:53,880
Speaker 2: degree of complexity, of operational complexity, of technological energy complexity

855
00:43:53,880 --> 00:43:55,840
Speaker 2: for these it just seems like exponentially higher.

856
00:43:55,920 --> 00:43:58,480
Speaker 3: Yeah, And I keep thinking back to just the sheer

857
00:43:58,480 --> 00:44:01,320
Speaker 3: scale of it, and like thebers that got thrown around

858
00:44:01,320 --> 00:44:04,520
Speaker 3: literally trillions of dollars. In the next few years. Well,

859
00:44:05,000 --> 00:44:06,640
Speaker 3: I'm sure we'll do more episodes on it.

860
00:44:06,680 --> 00:44:09,680
Speaker 2: There are so many sub episodes we could do, including

861
00:44:10,080 --> 00:44:12,959
Speaker 2: how the best of the Texas is ninety eight water

862
00:44:13,040 --> 00:44:14,280
Speaker 2: districts and how you find.

863
00:44:14,040 --> 00:44:16,400
Speaker 3: We No, no, no, Joe, you know what sub episode we

864
00:44:16,440 --> 00:44:16,839
Speaker 3: could do?

865
00:44:17,360 --> 00:44:20,120
Speaker 2: Subse cables we got to do. We didn't get into that,

866
00:44:20,160 --> 00:44:24,479
Speaker 2: but we should do an an Yeah, well we should

867
00:44:24,520 --> 00:44:27,640
Speaker 2: do more subs cable episode. And actually just in terms

868
00:44:27,680 --> 00:44:31,760
Speaker 2: of data center siting, like access to you know, latency

869
00:44:31,840 --> 00:44:34,399
Speaker 2: risk and where it needs to be, et cetera. It's

870
00:44:34,440 --> 00:44:35,520
Speaker 2: something we should talk about more.

871
00:44:35,640 --> 00:44:36,319
Speaker 3: Shall we leave it there?

872
00:44:36,360 --> 00:44:37,279
Speaker 4: For leave it there?

873
00:44:37,440 --> 00:44:39,760
Speaker 3: This has been another episode of the Odd Thoughts podcast.

874
00:44:39,880 --> 00:44:42,759
Speaker 3: I'm Tracy Alloway. You can follow me at Tracy Alloway and.

875
00:44:42,680 --> 00:44:45,200
Speaker 2: I'm joll Wisenthal. You can follow me at the Stalwart.

876
00:44:45,400 --> 00:44:49,000
Speaker 2: Follow our producers Kerman Rodriguez at Carman Arman Dashel Bennett

877
00:44:49,000 --> 00:44:51,520
Speaker 2: at Dashbot and kill Brooks at kill Brooks. For more

878
00:44:51,560 --> 00:44:54,040
Speaker 2: odd Laws content, go to Bloomberg dot com slash odd

879
00:44:54,040 --> 00:44:56,759
Speaker 2: Lots with the daily newsletter and all of our episodes

880
00:44:56,920 --> 00:44:58,920
Speaker 2: and you can shed about all these topics twenty four

881
00:44:58,920 --> 00:45:02,799
Speaker 2: to seven in our just discord dot gg slash online.

882
00:45:03,040 --> 00:45:05,880
Speaker 3: And if you enjoyed this conversation, if you want us

883
00:45:05,920 --> 00:45:09,080
Speaker 3: to do a sub episode on subc cables, then please

884
00:45:09,160 --> 00:45:12,040
Speaker 3: leave us a positive review on your favorite podcast platform.

885
00:45:12,080 --> 00:45:14,560
Speaker 3: And remember, if you are a Bloomberg subscriber, you can

886
00:45:14,600 --> 00:45:17,719
Speaker 3: listen to all of our episodes absolutely ad free. All

887
00:45:17,719 --> 00:45:20,040
Speaker 3: you need to do is find the Bloomberg channel on

888
00:45:20,120 --> 00:45:23,640
Speaker 3: Apple Podcasts and follow the instructions there. Thanks for listening