00:00There was a really interesting line in your notes. You talk about how the chip cycle and the
00:04hyperscaler cycle, they're beginning to diverge here. We've certainly seen that when it comes to
00:09the equity market performance as well. And you write that the more important question here is
00:14whether hyperscalers will pull back on their spending as the AI build out becomes increasingly
00:20costly. And that's exactly the question I want to ask you, because that's pretty existential when
00:25you think about this AI trade overall. Yes. I mean, I think that we are at the moment that we
00:30kind of
00:31witnessed the nature of the game is changing over the last five, six years. It has been a race for
00:38scale. So who can get to the kind of the scale first? Who can rule out the application first?
00:45But now they're paying attention to efficiency. So now the race is becoming like a race for
00:50efficiency. Who can do it at the reasonable ROI? And where does that kind of return on investment
00:56is coming is going to come from as a kind of question that we're kind of starting to ask.
01:02And it's interesting because, I mean, you know, you mentioned the race for efficiency,
01:06you know, a lot of scrutiny when it comes to ROI. I mean, you put it together. It sounds like
01:11just
01:11necessarily spending a ton on CapEx, funneling a whole bunch of money into this build out. That
01:18isn't necessarily the winning formula anymore. That's right. Absolutely. I mean, I think that's
01:25that's the point that I wanted to emphasize that over the last five, six years, it has been a game
01:33of capital. So like who can pour more capital at speed seem to be winning the race. What hasn't
01:41changed is AI is a transformative technology, and it's going to be like weaved into the fabric of our
01:47society in the coming decades. However, how it's going to be made is it's going to come from more
01:54kind of achieve more efficiency and putting more engineering effort to make it work better with a
02:00lower cost. Imagine what it was like when the when the mobile phone was first introduced, that there
02:07was a phase of the kind of the proof of concept like that. I mean, I think some of us
02:13still remember what
02:14car phone looked like. It took up one third of the space within the car. It was a great proof
02:20of concept.
02:21But if we can overbuilt ecosystem around that very product, it would have become obsolete very quickly,
02:29because it took decades from what it was like when in the car phone age to iPhone in our pocket.
02:36There's a lot of engineering innovation that took place from making that product to becoming the
02:44hands of everyone. That's what's happening with AI. The intelligent cost of intelligence should go down
02:50to make it more affordable and accessible to greater greater population around the world. And I think a lot
02:56of engineering innovation will come come out. And we're not certain if today's data centers and
03:03infrastructure the way it is is going to be suitable for that kind of future demand. Well, that's what I'm
03:08curious about. Where are you seeing right now in the early stages that type of engineering that might
03:14actually end up being the real realized future of AI? Yes. I mean, I see like a lot of the
03:19early stage
03:19company. That's why I'm very excited to work with this early stage founders who are solving the very
03:25engineering problems, how we can make compute more efficient, how we can make the like AI and
03:31intelligence affordable at like a lower cost, how we can optimize in terms of the kind of amount of data
03:38needed, amount of compute needed to get to the similar level of intelligence that we see today. I think
03:43those are the all open questions and a lot of the founders are kind of racing to get get there.
03:47Is that what when we look at and we don't have to mention specific companies, but when we look at
03:51the big
03:52companies that now are sort of leading this charge, do you think they're trying to work towards that? Or are
03:58they
03:58more just focus on whatever the sort of land grab is, for lack of a better phrase? I'm sure that
04:04like a lot of
04:05people realize that it's kind of that there is a they're turning tide that we have to also focus on
04:10like a bottom line cost and efficiency instead of just kind of showing that technology demonstration.
04:18So I think there is a lot of effort going into it. But always, I mean, if you look at
04:22the history of the
04:23innovation, they're like the small companies are agile and a better position to kind of come up with it, come
04:29up
04:29with the new solutions that completely kind of transform transformative ones because they don't have the
04:35baggage and legacy. So do you see the potential in the in the near term for consumer facing products? So
04:42more than just
04:43the chatbots for us to ask questions, actually being available and more importantly, actually being useful? Of course. Yes. Yeah.
04:50I think it's a kind of the AI,
04:52chatbot is not is not the only AI. I mean, I think we have to see this revolution as a
04:59more kind of adoption of the more software that what what you have we have to recognize is kind of
05:06that the cost for producing software has come down significantly. So in the industries and businesses which could not benefit
05:12from the advance of like availability of data and a lot of software will benefit from it will achieve the
05:19level of automation,
05:20quality improvement and efficiency. So I think those are the things that we'll expect. We we can expect to see
05:28as the AI development and innovation continues.
05:32And I want to situate what we're talking about in the macroeconomic backdrop, because, you know, this innovation that we're
05:38talking about,
05:40it's not happening in a vacuum. So I mean, you think about where we sit right now, we're talking about
05:45higher for longer interest rates, the idea of interest rates and thus the cost of capital going higher. How is
05:53that impacting the conversations that you're having with founders today, if at all? I mean, I think that I mean,
05:59another aspect of like working with early
06:02stage founders is that we're reminded every day that innovation and creativity comes from scarcity. So kind of those are
06:12trained. Those are the ones who are trained to work with the very limited resources. So in the backdrop of
06:18like kind of macroeconomic volatility, I think that they are trying to find the principled approach, finding the solution and
06:29breakthrough through the innovation and engineering effort, rather than
06:32than pouring a lot of capital into that.
Comments