WEBVTT - Tribe Capital Chairman and Co-Founder Arjun Sethi Talks AI

0:00:00.160 --> 0:00:03.840
<v Speaker 1>Joining us with more on Everything AI is longtime tech

0:00:04.000 --> 0:00:06.880
<v Speaker 1>entrepreneur Arjunseeti. He is the chairman and co founder of

0:00:07.000 --> 0:00:10.080
<v Speaker 1>VC firm Tribe Capital. He is also the co CEO

0:00:10.200 --> 0:00:14.000
<v Speaker 1>of Termina AI. Thanks for joining us, especially because your

0:00:14.080 --> 0:00:17.280
<v Speaker 1>firm has invested in Xai, so you have a very

0:00:17.320 --> 0:00:20.959
<v Speaker 1>close view on how these lms have been developing and

0:00:21.040 --> 0:00:24.439
<v Speaker 1>the competition among them. Very curious about what you think

0:00:24.480 --> 0:00:27.280
<v Speaker 1>about this latest open ai funding round and the corporate

0:00:27.320 --> 0:00:28.080
<v Speaker 1>interest around it.

0:00:29.080 --> 0:00:31.400
<v Speaker 2>Yeah, thanks for having me. I think the right way

0:00:31.440 --> 0:00:35.080
<v Speaker 2>to think about it is the corporate interest, especially from

0:00:35.720 --> 0:00:38.479
<v Speaker 2>a company like Nvidia, is more about if you take

0:00:38.479 --> 0:00:41.720
<v Speaker 2>a look at their portfolio differcification of where the revenue

0:00:41.760 --> 0:00:44.080
<v Speaker 2>comes from. The fastest growth in the future is going

0:00:44.120 --> 0:00:46.879
<v Speaker 2>to come from companies like OpenAI and the types of

0:00:46.960 --> 0:00:48.960
<v Speaker 2>developers that they're going to support moving forward.

0:00:49.440 --> 0:00:52.400
<v Speaker 1>So go ahead, please good.

0:00:53.040 --> 0:00:55.880
<v Speaker 2>So what that means is that Nvidia is not just

0:00:56.040 --> 0:00:59.200
<v Speaker 2>a hardware and a company. It's a hardware and software

0:00:59.200 --> 0:01:02.960
<v Speaker 2>company is attracting developers and they have to partner with

0:01:03.040 --> 0:01:05.280
<v Speaker 2>other developers and other companies similar to Opening I. So

0:01:05.319 --> 0:01:08.680
<v Speaker 2>Opening I being the biggest, it's a foregone conclusion that

0:01:08.720 --> 0:01:10.800
<v Speaker 2>they have to partner with them in some capacity similar

0:01:10.880 --> 0:01:11.279
<v Speaker 2>to Apple.

0:01:12.200 --> 0:01:16.040
<v Speaker 1>So given that you have invested in XAI, draw me

0:01:16.080 --> 0:01:19.479
<v Speaker 1>a map a little bit on where different companies are

0:01:19.520 --> 0:01:21.800
<v Speaker 1>going to play in this ecosystem.

0:01:23.000 --> 0:01:26.319
<v Speaker 2>So right now we're talking about llms, we're talking about

0:01:26.360 --> 0:01:29.160
<v Speaker 2>foundational models, but we're not talking about is similar to

0:01:29.160 --> 0:01:32.440
<v Speaker 2>the past, where you've built the foundation, you've built the infrastructure,

0:01:33.120 --> 0:01:35.360
<v Speaker 2>and the next layer of application companies are being built,

0:01:35.360 --> 0:01:38.000
<v Speaker 2>which is essentially what are the next trillion dollar companies

0:01:38.040 --> 0:01:40.959
<v Speaker 2>that are out there? So opening I being one foundational,

0:01:41.160 --> 0:01:43.759
<v Speaker 2>XAI could be another one that's foundational, Andthropic is another

0:01:43.760 --> 0:01:46.800
<v Speaker 2>one that's foundational, but they're supporting the next level of developers.

0:01:47.520 --> 0:01:50.360
<v Speaker 2>You take a look at what companies are doing today

0:01:50.440 --> 0:01:52.560
<v Speaker 2>similar to what a lot of venture firms are trying

0:01:52.560 --> 0:01:54.360
<v Speaker 2>to figure out, which is how do you leverage AI

0:01:54.480 --> 0:01:56.400
<v Speaker 2>to be more efficient or how do I say cost

0:01:56.520 --> 0:02:01.720
<v Speaker 2>or increase revenue? Today with the aspect of what you

0:02:01.760 --> 0:02:05.920
<v Speaker 2>see with open AI is that every single company, every

0:02:05.960 --> 0:02:09.280
<v Speaker 2>single company in our portfolio is leveraging these prompts and

0:02:09.320 --> 0:02:11.240
<v Speaker 2>leveraging AI in some way, and so what is in

0:02:11.240 --> 0:02:13.520
<v Speaker 2>net effect is that they're being twenty five to fifty

0:02:13.520 --> 0:02:16.040
<v Speaker 2>percent more efficient. It's not ten x yet. We're working

0:02:16.040 --> 0:02:18.880
<v Speaker 2>our way there as these companies get larger and they

0:02:18.919 --> 0:02:23.560
<v Speaker 2>build better developer tools, but that's an increased efficiency and

0:02:23.600 --> 0:02:27.120
<v Speaker 2>that has a high amount of impact. So any corporate

0:02:27.160 --> 0:02:29.640
<v Speaker 2>interest that's coming into open ai is looking at that

0:02:29.800 --> 0:02:32.080
<v Speaker 2>not just for themselves, but they're looking at that for

0:02:32.200 --> 0:02:34.920
<v Speaker 2>the future of their business, keeping the business that they

0:02:35.000 --> 0:02:37.680
<v Speaker 2>have with their developer community, and making sure that more

0:02:37.960 --> 0:02:41.440
<v Speaker 2>of these developers are going to continue to succeed on

0:02:41.480 --> 0:02:44.000
<v Speaker 2>their platforms because that's what's going to continue to drive

0:02:44.000 --> 0:02:46.440
<v Speaker 2>their business. So for Nvidia, it's a lock in because

0:02:46.440 --> 0:02:49.240
<v Speaker 2>they have software plus hardware, and that is the defect

0:02:49.280 --> 0:02:50.120
<v Speaker 2>of monopoly today.

0:02:50.800 --> 0:02:52.480
<v Speaker 1>Argene, can you talk to me a little bit about

0:02:52.520 --> 0:02:55.840
<v Speaker 1>valuation here? You have a report out from your company

0:02:55.880 --> 0:02:58.560
<v Speaker 1>about what you call a weather report about the supply

0:02:58.680 --> 0:03:01.400
<v Speaker 1>dynam and dynamics in price markets. When we talk about

0:03:01.400 --> 0:03:04.360
<v Speaker 1>open ai, we're talking about evaluation that could be over

0:03:04.400 --> 0:03:06.840
<v Speaker 1>one hundred billion dollars. But really there are only a

0:03:06.880 --> 0:03:10.079
<v Speaker 1>handful of companies in the world with that kind of

0:03:10.400 --> 0:03:13.720
<v Speaker 1>valuation underpinning them. So what else are you seeing in

0:03:13.760 --> 0:03:15.160
<v Speaker 1>the private markets right now?

0:03:16.760 --> 0:03:19.280
<v Speaker 2>So you know the way to start to think about

0:03:19.320 --> 0:03:24.760
<v Speaker 2>private markets is worldwide. So this take China, India, US,

0:03:24.800 --> 0:03:27.680
<v Speaker 2>and then Latin America as a whole. US and China

0:03:27.800 --> 0:03:30.240
<v Speaker 2>sort of dominated valuations for a while, and a lot

0:03:30.280 --> 0:03:32.760
<v Speaker 2>of capital has been flowing into anything that's software and

0:03:32.800 --> 0:03:36.160
<v Speaker 2>tech enabled. The next phase of that was anything that

0:03:36.320 --> 0:03:39.240
<v Speaker 2>software tech enabled and the label of machine learning and AI,

0:03:39.840 --> 0:03:41.480
<v Speaker 2>and so that's where a lot of the capital is going.

0:03:41.680 --> 0:03:45.800
<v Speaker 2>So far, most of the investments in the private world

0:03:45.920 --> 0:03:49.840
<v Speaker 2>that's been going into AI has been coming from corporates.

0:03:49.880 --> 0:03:52.960
<v Speaker 2>It hasn't been coming from venture capital. Venture capital traditionally

0:03:53.040 --> 0:03:56.640
<v Speaker 2>is invested in anything that's vertically integrated or vertical application,

0:03:56.960 --> 0:04:00.240
<v Speaker 2>which basically means like how do you leverage AI into

0:04:01.520 --> 0:04:04.160
<v Speaker 2>a fintech, how do you leverage AI into healthcare, etc. Etc.

0:04:05.120 --> 0:04:07.040
<v Speaker 2>Those are where most of the investments are going in.

0:04:07.120 --> 0:04:11.120
<v Speaker 2>So valuations for those types of opportunities in the beginning

0:04:11.120 --> 0:04:13.800
<v Speaker 2>are really high. Then midway through the cycle it becomes

0:04:13.840 --> 0:04:16.040
<v Speaker 2>low again, and then what you'll see is something that's

0:04:16.080 --> 0:04:18.920
<v Speaker 2>much more intrinsic versus options value related.

0:04:19.800 --> 0:04:22.160
<v Speaker 1>You know, when you think about the global view here,

0:04:22.240 --> 0:04:24.760
<v Speaker 1>there's a lot of concerns in the AI world about

0:04:24.760 --> 0:04:29.080
<v Speaker 1>that competition towards AI spending towards AI development between the

0:04:29.160 --> 0:04:34.000
<v Speaker 1>US and China. What do the dollars say, Well, most.

0:04:33.800 --> 0:04:37.640
<v Speaker 2>Of the dollars today are going towards development in the

0:04:37.720 --> 0:04:41.520
<v Speaker 2>United States. That's actually very very clear. That said, what

0:04:41.600 --> 0:04:44.600
<v Speaker 2>you have to look at in terms of what investments

0:04:44.600 --> 0:04:47.680
<v Speaker 2>had happened outside of the United States, especially in China,

0:04:47.760 --> 0:04:53.120
<v Speaker 2>was essentially for image recognition, anything that was related to

0:04:53.240 --> 0:04:56.360
<v Speaker 2>cybersecurity or surveillance. And so that's how they had spent

0:04:56.560 --> 0:04:58.919
<v Speaker 2>majority of their capital, let's call over the last five years,

0:04:59.240 --> 0:05:02.840
<v Speaker 2>and we spent that plus more over here that's been

0:05:02.880 --> 0:05:07.320
<v Speaker 2>around AI, AGI machine learning. So we're much further ahead

0:05:07.720 --> 0:05:09.520
<v Speaker 2>for now. And you can see that in the types

0:05:09.520 --> 0:05:11.320
<v Speaker 2>of companies that are being built, types of products that

0:05:11.360 --> 0:05:16.520
<v Speaker 2>are being built for every single vertical. So I always

0:05:16.560 --> 0:05:20.320
<v Speaker 2>go back to cybersecurity, healthcare, financial services. You're going to

0:05:20.320 --> 0:05:22.960
<v Speaker 2>see the first input there. You could call it GDP

0:05:23.160 --> 0:05:27.039
<v Speaker 2>per capital growth and then cost. So those are the

0:05:27.040 --> 0:05:29.680
<v Speaker 2>bifurcations between the investments that are really being made and

0:05:29.720 --> 0:05:32.440
<v Speaker 2>then all the hyperscalers that are out there. They are

0:05:32.480 --> 0:05:34.240
<v Speaker 2>going to benefit from all of this because you need

0:05:34.279 --> 0:05:36.679
<v Speaker 2>more compute, you need more space, you need more training.

0:05:36.720 --> 0:05:40.000
<v Speaker 2>You need more inference, which ends up being in net

0:05:40.000 --> 0:05:42.159
<v Speaker 2>benefit for all the companies that are part of that stack.

0:05:42.400 --> 0:05:44.920
<v Speaker 1>Speaking of hyperscalers, I'm glad you brought this up. Of course,

0:05:44.920 --> 0:05:48.600
<v Speaker 1>we had Nvidea earnings this week. The initial reaction was disappointment.

0:05:48.839 --> 0:05:51.320
<v Speaker 1>But something that struck me on the heels of Unvideo

0:05:51.400 --> 0:05:54.520
<v Speaker 1>earnings is, even when you saw a day of earnings

0:05:54.520 --> 0:05:57.599
<v Speaker 1>where the stock was down right after, you saw every

0:05:57.680 --> 0:06:02.000
<v Speaker 1>other company in the Philadelphia Semi Conductor Index immediately have

0:06:02.120 --> 0:06:05.160
<v Speaker 1>the reaction in the opposite direction, there is still love

0:06:05.240 --> 0:06:09.280
<v Speaker 1>for that AI boom. Is this argent because there is

0:06:09.400 --> 0:06:12.440
<v Speaker 1>such a lack of investment opportunity. You mentioned a little

0:06:12.440 --> 0:06:14.599
<v Speaker 1>earlier that a lot of this benure investment is actually

0:06:14.600 --> 0:06:18.159
<v Speaker 1>coming from corporations. So what does that mean in terms

0:06:18.160 --> 0:06:21.480
<v Speaker 1>of exit opportunities? Are there fewer in the future than

0:06:21.520 --> 0:06:22.360
<v Speaker 1>meets the eye?

0:06:24.040 --> 0:06:27.159
<v Speaker 2>Yeah, so there's multiple questions there. I'll take a step back.

0:06:27.320 --> 0:06:31.200
<v Speaker 2>So you look at Nvidia, and the way in which

0:06:31.200 --> 0:06:33.480
<v Speaker 2>you should think about it from our framework is that

0:06:33.520 --> 0:06:36.359
<v Speaker 2>they've have clear product market fit for hardware that they've built,

0:06:36.560 --> 0:06:38.640
<v Speaker 2>and they have clear product market for the software that

0:06:38.720 --> 0:06:41.560
<v Speaker 2>enables their hardware. A lot of the other companies that

0:06:41.600 --> 0:06:44.640
<v Speaker 2>compete with them today don't have that, and so they're

0:06:44.720 --> 0:06:49.440
<v Speaker 2>racing to be able to lock in developer interests. So

0:06:50.080 --> 0:06:54.560
<v Speaker 2>all of them, they're racing to be able to try

0:06:54.560 --> 0:06:57.599
<v Speaker 2>to commoditize that part of the stat stack that hasn't

0:06:57.600 --> 0:06:59.760
<v Speaker 2>happened yet. And I think that's a key point. So

0:07:00.240 --> 0:07:02.360
<v Speaker 2>when you think about year over year growth of what

0:07:02.480 --> 0:07:04.920
<v Speaker 2>Nvidia looks like, if you think about year over year

0:07:04.960 --> 0:07:06.880
<v Speaker 2>growth for the other companies that are on top of

0:07:06.880 --> 0:07:10.880
<v Speaker 2>their stack, you know, namely open Ai, Xai Andthropic, et cetera.

0:07:11.400 --> 0:07:13.000
<v Speaker 2>All of these companies have to rely on that, and

0:07:13.000 --> 0:07:14.240
<v Speaker 2>then you're going to look at the next part of

0:07:14.280 --> 0:07:16.400
<v Speaker 2>the stack, which is what do all the developers do,

0:07:17.200 --> 0:07:18.760
<v Speaker 2>how do they train, and what are the products that

0:07:18.800 --> 0:07:21.080
<v Speaker 2>they're going to use. Beyond just open source, they still

0:07:21.120 --> 0:07:24.760
<v Speaker 2>have to work off of hardware. So I think of

0:07:24.800 --> 0:07:27.880
<v Speaker 2>this very similar to when Apple came out. Everyone had

0:07:27.920 --> 0:07:30.679
<v Speaker 2>made a bet that you know, Apple's devices are too expensive,

0:07:30.680 --> 0:07:32.360
<v Speaker 2>it's going to be commoditized and all of the other

0:07:32.360 --> 0:07:33.960
<v Speaker 2>people are going to come in and compete with them,

0:07:34.680 --> 0:07:38.240
<v Speaker 2>and then you have software developers moving over to Android.

0:07:38.240 --> 0:07:40.480
<v Speaker 2>You have roughly about a fifty to fifty market. You

0:07:40.480 --> 0:07:43.320
<v Speaker 2>don't have that today. You don't have a clear competitor

0:07:44.000 --> 0:07:46.720
<v Speaker 2>in hardware, and you don't have a clear competitor in software.

0:07:46.800 --> 0:07:48.880
<v Speaker 2>So the question you have to ask is that what

0:07:49.080 --> 0:07:53.160
<v Speaker 2>point does that start diverging in terms of overall demand

0:07:53.240 --> 0:07:56.400
<v Speaker 2>for processing and demand for inference and demand for training.

0:07:56.960 --> 0:07:59.120
<v Speaker 2>And today it hasn't stopped. I do believe it will

0:07:59.440 --> 0:08:02.680
<v Speaker 2>will sort of now level out at some point, but

0:08:02.760 --> 0:08:04.720
<v Speaker 2>we don't know if that's six months from now or

0:08:04.720 --> 0:08:06.720
<v Speaker 2>two years from now or ten years from now.

0:08:07.600 --> 0:08:09.480
<v Speaker 1>Rgine, we have to leave it there. That is Termina

0:08:09.560 --> 0:08:13.080
<v Speaker 1>AI co CEO, Rgine Sethie. We appreciate having you today.

0:08:13.160 --> 0:08:14.600
<v Speaker 1>Have a great long weekend.