00:00 Speaker A

Do you think that publicly traded AI is in a bubble?

00:03 Speaker B

So, what it’s an interesting question in that, you know, you have to build out infrastructure. and again, being a student of history around as an investor, you have to be, it took massive amounts of capital to build out the infrastructure that we benefit from now. You remember the global crossings of the world building out cable and and cable and and fiber and okay, now and some of those companies went bust, there was in a bubble, some of those companies ended up being the winners because they built at the right time, right capacity in the needs and demands of what the market was demanding.

00:31 Speaker A

Although in some cases they took decades to come back in terms of the stock prices.

00:35 Speaker B

In some cases, absolutely. So that’s those are the dynamics. Everyone looks to be, we’re putting trillions of dollars into an infrastructure that is a question of do you have enough power to support it? You you argue is there is there a bubble because they can’t support it with the power. There’s an argument around that. Are there enough component parts? You know, is there enough distribution? Are those component parts that that we are looking at today going to actually have the give the power demands that are required. We’re hearing, of course, there are other solution sets that have lower power by a 10th of what we’re seeing. So do you need the same sort of power? So all of those are are market dynamics that have to reach certain parts of equilibrium over time. Where we focus, again, all of that provides infrastructure for enterprise software. At the end of the day, what we’re going to be doing is utilizing inference. You can put it in the form of tokens, utilizing inference to create massive economic rent and productivity for our customers.

01:29 Speaker B

That will all inure to the benefit of those enterprise software vendors who become agentic. So there’re going to be ebbs and flows over who is actually capturing what value at what period of time. Certain folks will have monopolistic positions for short periods of time and then others will come up with like in all Silicon, you know, different solution sets that will be lower cost, suit to purpose, different infrastructure. Do you have to have it all centralized? Can you do it at the edge? You know, in terms of the inference capabilities? Can you do do you need, you know, one gigawatt power supply or do you need, you know, 50 megawatts to do certain inference for certain use case applications? And is that, is that more suitable for that use case application. So we’re going to go through those waves and some parts of the this market are in bubbles and other parts are very much in a great investible space.

02:22 Speaker A

Do you think that power is going to be the main constraint on the build out of AI in the shorter term?

02:30 Speaker B

In the short term in the US, yes. I think in other parts of the global economy, they have an excess of power, relatively speaking. The question is can they utilize that power to actually power uh, Gen AI or agentic or or uh uh uh data systems or GPU data centers in a way that is accessible and usable for those who need it. So again, there’s there’s not only the power dynamic, there’s accessibility, the, you know, what are the requirements of latency and what are, you know, the sizes of the models for the use case and application. Not every model needs, you know, a a a massive LLM, some will be able to use some small language model in an in an environment to deliver a high capacity solution for that particular industry. So,

03:00 Speaker B

again, it’s early, it’s a lot, uh, but it has massive implications and massive productivity. So we’re excited to be a part of what I call what they the the edge of this thing, which ultimately is the enterprise software where we’ll capture benefits. We’ve built these businesses now. We’re seeing, we’ve we’re converting, I think we have 18 of our companies now that already have agentic solutions in the market that are getting installed or have been installed and seeing real results. and I am just really excited. I’ve never been more excited about investing in enterprise software than I am today.