Without Applications, there is no AI

What happened

Last week was a tough one for the software sector, which experienced a significant correction. This pressure culminated on 30 May with a large hit on the entire sector in the wake of Salesforce Inc’s results. Behemoths like ServiceNow Inc or CrowdStrike experienced double-digit dips during the week, and the bleeding spilled into many other well-known names (e.g. Nutanix Inc), muting into a broad-based software sector sell-off. Although the main trigger was the earnings season, the shadow of AI and its unexpected effects was omnipresent. 

Impact on our Investment Case

Earnings, what else? 

If the sell-off was triggered by Salesforce Inc, they were not the only ones impacted by the apparently sudden expected earnings slowdown. Across the software segment, companies have been cutting their guidance for the year, with the leitmotiv being a weakening macro. However, although the macro deterioration in the U.S. is apparent in recent GDP numbers, PMIs are not yet so negative – for now. And the deteriorating macro did not prevent major cloud players from posting excellent numbers. 

More interesting, the magnitude of the guidance cut is not that huge (e.g., Mongodb Inc cutting its FY25 guidance by ~200bp), but for a few players hit by company-specific execution problems (e.g., UiPath Inc) which do not call the broader business perspectives into question. Put bluntly, if the earnings season is not flamboyant, it is not a disaster either. We agree that the valuations of prementioned such companies leave little margin for error, but the market reaction would appear out of proportion. Something else is at work. 

Well, actually... 

This something else is AI, but not as we would like, nor as we had anticipated. It is becoming apparent, to some extent, that AI has become such a prominent topic for executives that they have simply decided to sit on the sidelines. Many companies indeed invested massively in their software infrastructure following the pandemic. They are now facing a situation where this infrastructure is possibly already obsolete, and where they lack clarity of vision about how to continue investing, as the technology panorama is evolving at a pace rarely seen. Hence a pause to assess the direction of further investments, leading to the inevitable company comments of “deal postponements” and “additional scrutiny”. 

Another interesting point is that there seems to be no massive deployment of generative AI applications yet – coding excepted. This should not come as a surprise, as the technology is in its early phase. Companies are slowly moving out of the proof-of-concept stage, but the transition towards the mass deployment phase cannot be immediate nor smooth. IT departments indeed need to rethink and adjust their IT strategies, considering for example how effective would be implementing tools such as GitHub’s Copilot in their current iteration, i.e., something with the potential to leverage the capabilities of a skilled developer but to ruin an infrastructure if left in unexperienced hands. This dilemma is obviously affecting planning and decision making in the short term. 

What is the market trying to tell us? 

Beyond these effects, which we believe are not bound to last (and could not justify such a market movement), we think there is a broader calculus ongoing. Many of the companies that experienced the bulk of the sell-off could be considered as sensitive to AI displacement. Salesforce Inc is a textbook example. Despite being a prominent apologist of AI deployment, its technological stack is fatally behind the curve compared to newer players: “AI native” players will be more capable of proposing advanced features for salespeople, such as call prioritization, deal review, strategic hints, or more accurate forecasts because such technologies will be built-in features, while Salesforce Inc’s solutions do little to none of this today. This puts a potential ceiling on future growth.  

More important, in a world where AI technologies can generate code on-the-go from a simple prompt, barriers to entry are shrinking and competitive advantages are dissolving. Therefore, the market is now suddenly awakening to the risks related to “legacy” software players. Although we agree that the risk exists over the longer term, we do not think that replacing the backbone of your IT system can be done overnight: AI or not, no sane CEO would risk paralyzing its business for purely technical reasons, and the transition would be gradual, leaving ample time to adjust course. In addition, we do not agree with the fact that every company having experienced a dip can be easily replaced: for every Salesforce Inc, there is a Datadog Inc, which sits at the core of the cloud infrastructure and already mechanically benefits from the rise in AI workloads.  

What’s next?

Such a transition would however lead to the question of who the winner will be. We’ve always stated that picking a winner among applications was an uncertain business, especially in the early phase of a technology transition, hence our focus on the earlier stages of the supply chain. This uncertainty rhymes  with lower general valuations, with only the few champions-of-the-day remaining unscathed (greetings Microsoft). Time will tell if we’ll see a winner-takes-all scenario, the emergence of an oligopoly, or a large, diversified ecosystem.  

What we cannot leave aside is the strength of hardware players, although Marvell ’s results showed that they are not totally immune to the pressure. NVIDIA Corp, in particular, has benefited from a shopping spree from large players to build up capacity, and the spread between hardware and software valuations has reached historical lows. 

 

However, should the high expected demand not materialize (which is not our scenario), either for macro reasons or because deployment keeps being postponed, then the hardware sector would be at risk of a rude awakening (hence making our positioning on software players even more compelling). Because no applications would simply mean no AI – both as a technology or as an investment story. 

Our Takeaway

Our underlying investment thesis of software becoming increasingly significant in the AI investment theme remains unshaken. With the shift towards the application phase now unfolding, we think the best returns for our AI & Robotics strategy are still ahead. Although we acknowledge that we had not foreseen such transition pains, they paradoxically reinforce our conviction regarding the importance of AI in businesses, considering how companies appear focused on managing a successful deployment. We are convinced that the bulk of our portfolio is on the good side of the fence. We will adjust it to limit our exposure to companies with the higher execution risk, and will take the opportunity to add on stocks such as Mongodb Inc - the savvy investor will take the opportunity to enter once the froth is shaken out.