It’s been a tough H1 for software stocks, which have lagged hardware/semiconductors names by 20-30% on average. While expectations built up last year about the revenue opportunity for many software companies shipping new products with built-in AI, the reality caught up during the latest earnings season with the likes of Salesforce, MongoDB, Workday, Cloudflare and Rapid7 delivering underwhelming quarterly figures and/or guidance due to a weak software spending environment.
Surprisingly, there was no consistent message to explain this weakness. If we can assume that some enterprise customers kept digesting their massive software spending of 2020-2022, we also believe that AI has been until now more a headwind than the expected tailwind for two main reasons.
First, it’s highly likely that enterprise customers have started to divert some of their IT budgets towards in-house AI initiatives at the expense of “traditional” software spending. Even if this software comes with AI features, company CTOs in most industries are now prioritizing the development and roll out of their own AI products and services.
Second, it’s also highly likely that CTOs at enterprise customers are reluctant to commit to new software given the flurry of AI software products hitting the market right now, the time to review them and the rapid pace of AI advances that could make many of these first-generation AI software obsolete in one or two years. In short, we suspect that customers will remain in a wait-and-see mood in the very short term.
In our view, these headwinds could dissipate in the next couple of years as the technology matures and customer CTOs become more comfortable with it. Given the current negative sentiment on the software sector, any good set of quarterly earnings at some point could then restore confidence and spark a rebound. That said, longer term AI challenges are emerging and posing fundamental risks to entire segments of the software industry, suggesting that one will need to be very careful going forward when investing in the sector.
A few days ago, Bloomberg reported that Google, which has merged one year ago its two AI labs Google Brain and DeepMind, has decided to transition the new unit from a research lab to an “AI product factory”, in other words a software company selling products/services. This move obviously comes in response to similar commercial initiatives from other AI players such as OpenAI.
It turns out that DeepMind has already developed industry-specific models that could turn into revenue generators soon. We’re notably thinking of AlphaFold 3, an AI model capable of predicting the structure and interactions of biological molecules, expected to accelerate the drug development process.
DeepMind also unveiled in recent months its text-to-video games model, called Genie. Genie, which was trained on 30,000 hours of video of hundreds of 2D platform games, is able to build playable games in the style of classic 2D platformers like Super Mario based on a prompt that can take the form of a short description, a hand-drawn sketch, or a photo. We had already highlighted two months ago that Google DeepMind had made no secret that Genie, that is an in-house research project, could one day be turned into a commercial game-making tool, notably when it becomes able to produce 3D games/environments.
Overall, it’s becoming obvious that some of the most powerful AI models are going to compete head-to-head with traditional software solutions and we can then question the long-term outlook of companies such as Schrodinger (drug molecule design software) or Unity (video game design software). Worryingly, the capabilities of AI models will not only be limited to healthcare and video games and they could soon hit many other software segments. It’s worth noting here that Google is reportedly in talks to acquire HubSpot, a move that would open the CRM software market to the Tech giant, and put it in direct competition with Salesforce…
In conclusion, even if the AI initiatives of software vendors are likely to spark some topline upside in the next 12-24 months, we are concerned that the competitive environment in the software space is going to materially intensify in the mid to long term and keep pressuring valuations. Accordingly, we feel more comfortable being overweight hardware, also considering that inference and edge AI are going to open a new growth cycle for semiconductors.
