Transcript
Speaker: Welcome to HSBC Global Viewpoint, the podcast series that brings together business leaders and industry experts to explore the latest global insights, trends, and opportunities.
Speaker: Make sure you're subscribed to stay up to date with new episodes. Thanks for listening, and now onto today's show.
Speaker: This is the Macro Brief from HSBC Global Investment Research, the podcast that looks at the issues driving financial markets. And I'm Pierre Butler. In recent days, software stocks have been hit by a sell-off.
Speaker: The cause? Fears that the business models of software vendors could be upended by artificial intelligence. So on today's podcast, we're looking at whether these concerns are justified and asking, can AI replace software?
Speaker: To do that, I'm joined from Florida by our head of US technology research, Steve Bursey. Steve, great to have you on the podcast. Thank you, pleasure to be here. So I was going to start ah by quoting ah the father of value investing, Benjamin Graham. And the quote is, in the short term, the market is a voting machine and in the long term, it's a weighing machine. So in the case of the software stocks, it looks like the market has voted in the short term that their business models are going to be severely undermined by by AI. Maybe just kind of recap on what's been going on so far.
Speaker: Yeah, so for for the listeners, you know, there's been a dramatic pullback in a lot of the technology names and some concentration on the software side. So there's been some high profile media um ah articles and some chatter from market pundits about the demise of software. And what we quickly saw was somewhat of an echo chamber where that that information rapidly dispersed across the the markets. And we saw some what we think was indiscriminate selling. So to us, that that looks like a rush to just get me out.
Speaker: without really digging into the facts. And what I'd say about that is I can understand it. You know, if you're a portfolio manager, this is AI is one of the most technically complicated technologies that I've seen in my career, my 40 years in technology, 25 on the street. And the implications are, you know, you really have to dig in to what AI actually is, what software actually is, and is there a fit? you know Can you do this? Our opinion is that you know software is not on the enterprise class ah side is not an appropriate fit for LLMs and other competitors so simply using Vibe coding. The companies themselves are best suited. We did publish a report where recently where we dug into all of these factors. on why we think AI is not appropriate, but it is very complex and I can understand a rush to judgment.
Speaker: Steve, what do you mean by vibe coding, by the way? So vibe coding is in the headlines. It's the latest thing. It's where you use a very highly highly instructed agent that is a vibe coding agent. And it really works as a partner with you as a co-worker where you assign it various tasks and it will automatically create some code for you. But the key there is that you have to really understand what you're asking for because it will only give you really as much as you instructed to do. So the onus of the heavy lifting of getting it right is on the coder's shoulder in through those instructions.
Speaker: So to put it more simply, just making code is not enough. And in a way, people say, oh, you've written this code that can do this clever processing. And everybody panics about a company that has an established business model around this field. But there are several issues here on there. There's one of scaling and is also the other one, which you talk about in your report, is of reliability.
Speaker: That's right. So, you know, the world is used to deterministic systems. AI is undeterministic. So deterministic is, you know, coded, codified logic, business logic, where the result is we have a system with very little or no errors. We're used to no ever errors in our general ledgers and our day-to-day business operations. But AI, can we've all seen hallucinations. So there are very limited cases, and we talk about that in our report just published. And um there are limited cases, and the media has picked up on this, for instance, just a narrow-focused legal program.
Speaker: they there You can potentially replace a narrow focus program or image creation. A video editing program can be replaced with generative AI now. But those images and videos, you can have errors. You can have pixel errors in the background. As long as subjectively the eye thinks that it looks acceptable, that's fine. But no errors, I can tell you that, on a Swiss bank's general ledger are acceptable.
Speaker: And also the other issue, of course, which you mentioned your report, is ah the availability of the data that, say, an LLL model would work on. I mean, it's fine if the data is public, but if it is actually contained within an enterprise, that that's more of a challenge.
Speaker: That's right. So the limited cases I mentioned, both image and legal, the legal system and document and case laws are all publicly available on the Internet.
Speaker: If you want to try and train an LLM um to make images, there are billions of it, trillions of images publicly available on the internet. So if you try and replace those two, you can train on publicly available document. However, if you're looking at the most complex machines on the planet, enterprise class, high fidelity, high performance, high throughput, enterprise class software, None of that is public. It's it's highly, highly protected. And there's employees that move from one company to to the other. So the companies themselves make it very difficult to understand how they how they make their platforms and they design it for for high reliability.
Speaker: Is there also an issue that that people are saying, well, AI is making potentially complex procedures simpler or can be sort of realized far more quickly and that undermines the value proposition of of software? do you Do you believe that to be the case?
Speaker: Well, you know so that the cost of production is declining. So programmers are more productive. They're pumping out more code. So guess who is using the same Vibe coding tools that want to be competitors to the software industry are using? It's the software companies themselves. In fact, they're probably best suited to use Vibe coding to improve their software. So they're becoming more productive. They're not a target that's standing still. They're a moving target.
Speaker: And as far as the value creation goes, just because it's cheaper doesn't mean that customers will necessarily buy it at at a less expensive price point. The software model really is predicated on a value creation or a value generated model, a perceived value model where the customer and the seller both know the value of the solution that's being delivered. They can understand the value creation and that's what dictates the price being sold. It almost, software has almost a a zero marginal incremental cost.
Speaker: It's not like traditional hardware or say automobile cost plus models. It's a perceived value model. you You say quite an interesting example ah in your report in the context of Microsoft Office. Maybe tell us a little bit about that. Yeah, yeah. And a company I've covered for more than 25 years. So, yeah. So Office um was you know being speculated that it was the end of Office.
Speaker: because Google Docs and originally it was OpenOffice was being developed. I think it was under Sun Microsystems. There's a nick name from the past. And they were actually giving a competitor suite away for free. And of course, Google Docs followed in suit many years after. But even with those two applications available for free and plenty of functionality and team sharing ability and all of the like, um they've really not managed to make a dent in the enterprise footprint that Microsoft has with their Office suite.
Speaker: And on on that note, you you titled your report, Software is Still Alive and Set to Thrive. So not only are you saying that the recent market reactions perhaps are are overdone, but you actually see a very positive long-term scenario for software companies because of AI. So maybe just to finish on, explain that optimistic view.
Speaker: So the TAM, we see the TAM for the software sector, for the enterprise class software sector to really be expanding. And that really is being driven, in our view, by the by the use and the diffusion of AI within the enterprise users, the global two thousand s that will more and more as ai is diffused through these enterprise class systems for their day-to-day operations, that will make them much more efficient and they will be able to you you know utilize those efficiencies to lower their cost.
Speaker: Any efficiencies that are gained, they're going to be likely sharing that with the software vendors. That is their model. So the TAM of the global is really the efficiency gains of that 100 trillion plus USD global GDP and the manufacturers of them. So the efficiencies that can be gained in the production of global GDP are set to are really the TAM that I'm talking about.
Speaker: And for our listeners, TAM is the total addressable market. And your point is that that's likely to grow significantly in the future. That's right. And we're seeing, you know, we think that 2026 is the kickoff of that diffusion of AI throughout the enterprise class software stacks. It takes about two years to write the code. So while the market has been disappointed, where's the monetization? What we've been saying for two years is it takes two years to integrate it into these large platforms. So, yeah, so we think 2026 the kickoff you year for the monetization side. And if you look, you can really see the revenue streams for a lot of the vendors that we cover really accelerating and lifting from single digit top line growth of mature players up into the teens. And then the smaller, smaller scale vendors really getting into the high 20s and even thirty s
Speaker: On that note, we'll definitely have you back to see whether that is coming through. But that sounds like quite ah an optimistic forecast um relative to what's been happening in the market. So, Steve, thank you very much for joining us.
Speaker: Thank you. My pleasure.
Speaker: Before we go, a reminder that voting in the Extel Develop Europe and emerging EMEA surveys is about to close. If you're an HSBC client participating in them and you've valued our service and our insight over the past year, please consider supporting us.
Speaker: Head to extelinsights.com slash voting to cast your votes. But please be quick as voting ends on Friday, the 13th of February. Don't forget that you can keep up to date on all our latest reports and videos by downloading our app from Apple's App Store or Google Play.
Speaker: And you can listen to our Asia-focused sister podcast, Under the Banyan Tree, wherever you get your podcasts. And if you've got any questions or comments, then you can get in touch with us at askresearch at hspc.com.
Speaker: So that's it for today This week's podcast was hosted by me, Piers Butler, and produced by Tom Barton. Remember to like and subscribe to The Macrobrief wherever you get your podcasts.
Speaker: We'll be back next week.
Speaker: Thank you for joining us at HSBC Global Viewpoint. We hope you enjoyed the discussion. Make sure you're subscribed to stay up to date with new episodes.
