The software sell-off, part II

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Good morning. Treasuries have risen briskly all accross the yield curve in the past few weeks. The 10-year yield has fallen from 4.27 per cent to 4.06 per cent. Does the market think the economy is slowing, despite solid recent data? Is it responding to falling Japanese yields? Or are fears of AI deflation and job losses creeping into the bond market? Let us know what you think: unhedged@ft.com.

The software sell-off, part II

Last Friday, we took an initial look at the sell-off in software stocks, laying out which companies had been hit, by how much and when. But we didn’t make any progress on the question that really matters: how much of a threat do AI machines pose to the long-term profit growth of software companies? I know hardly anything about AI (a situation I’m trying to correct), so in taking a first shot at this question I’m depending on the arguments of others, a little logic and my experience of corporate finance.

The basic intuition behind the sell-off is that AI machines will allow companies and individuals to perform tasks they previously completed with paid-for software, or allow them to build substitutes for that software — all at a lower cost. Is this realistic? The consensus on Wall Street is some software vendors will be much more vulnerable than others and the sell-off has not respected the distinction between the endangered and the resilient.

Goldman Sachs recently proposed a software “pair trade” for clients — two buckets of stocks, one to buy and the other to short. They describe the longs as software companies that will be hard to replace because of “regulatory entrenchment, integration complexity, or human accountability” and that support “data infra[structure], observability, security, hyperscale cloud, [and] AI development platforms”. On the short side are companies supporting “workflows that AI could increasingly automate or rebuild internally”.

According to a Bloomberg story about the Goldman pair, stocks on the long side include Cloudflare, CrowdStrike, and Palo Alto Networks (all in cyber security), as well as Oracle and Microsoft. On the short side are Monday.com (application building, project management), Salesforce (customer relationship management), DocuSign (electronic signatures), Accenture (tech consulting) and Duolingo (language learning).

Software analysts, for their part, lean into the “integration complexity” argument for why software companies will continue to grow. The argument is based on high switching costs, the integration of different software systems and functions, and the ability of software companies to “plug” their own AI tools into the interconnected clusters of software their customers use. Here is HSBC’s Stephen Bersey:

The technology marketplace is an odd thing where the best or cheapest products don’t always win…

[O]nce a large platform application is installed and a customer’s business runs its critical operations on it, switching carries many risks . . . if there is a disruption during a platform replacement, normal operations can stall, loyal customers can leave forever, brands are tarnished and revenue can hard-stop until the issues are resolved . . . There are also many other unforeseen switching risks/costs that can fester, like lower productivity and numerous process errors as users retrain and get familiar with a new system. Or the sudden appearance of unwanted feature interactions that only manifest under long duration at-scale operations.

And here is Keith Weiss at Morgan Stanley:

[AI] risk should be considered within the framework of the end customer’s preference for ‘best of breed’ solutions versus more consolidated ‘suites’. In our [Chief Investment Officer] survey work . . . one of the driving factors of a rising preference for suites, in our view, [is] the ability of incumbent suite vendors to more effectively act as fast followers and minimise the functionality gap between their suite and best of breed solutions…

The direction of travel in software architectures over the past 50 years has been an increasing modularisation of the system into optimised components providing the best efficiency and efficiency for the type of work needing to be done. While the [AI can] effectively enable new functionalities around contextual understanding of unstructured data, coding, and content creation, [it is] unlikely to be the most effective and efficient component for every part of system

This argument makes sense as far as it goes: incumbent software systems are complex and hard to replace, and software companies will have been working hard on AI for some time. But how far does it go? The sell-off is not about damage to software companies’ profits in the next few years. Everyone acknowledges that industry fundamentals are fine and will continue to be fine for some years. This issue is how computing looks in a decade. In financial analysis terms, the question is about the “terminal valuation multiple” in the discounted cash flow model of your favourite software company. In a standard DCF model, you estimate profits for each of the next 10 years or so and then put a multiple — 20x, say — on the profits in year 10, resulting in a lump sum that accounts for all the future value. I think what is happening now is investors are taking that 20x and moving it to 15x or 10x because profits might not be growing very well at that point.

It’s not at all clear to me that this is a mistake. There are powerful network effects and high switching costs in the software business, which make it hard for new competitors to displace incumbents. But I’m not sure that AI machines are a new competitor in the standard sense. There may be no moment when customers take the old software system out and put the new AI system in. There may be no “switch”. Companies and individuals may just wake up one morning in five or 10 years and decide that, given the range of things their AI assistant can do for them, that they don’t need the old software any more. This is just a guess by a non-expert, but that’s what the threat look like to me.

A final point, one that several people have mentioned to be. The market currently acts as if the impact of AI will be sector specific, although which sector is expected to get hit changes from day to day. But if AI is revolutionary as many people think, won’t it be a huge, unpredictable “deflation bomb” that hits most sectors (I borrow the phrase from Paul Kedrosky)? That is to say, hasn’t the whole market gotten riskier, so the whole market needs to sell at a lower valuation? We may get there yet.

Good reads

Wild times at CBS.