The software sector has remained a minefield for investors this year, in line with our scenario (see our June 2024 report). So far, only two clear winners of the AI revolution have emerged—Microsoft and Palantir, both core holdings in our portfolios—while many other software vendors face the risk of being disrupted by AI.
That said, a select group of data science software vendors may finally be joining the ranks of beneficiaries. Snowflake (a portfolio company) and MongoDB, both focused on cloud databases and data engineering/preparation, reported impressive quarterly results this week, signaling a long-awaited inflection after years of elusive AI-related upside.
While data engineering—the collection, cleaning, transformation, and organization of data for AI and applications—is critical in the AI space, enterprise spending on AI software tools has lagged the massive spending by hyperscalers on AI model development, with growth at just 10–25% in recent quarters. This weakness largely reflects enterprise behavior: many companies opted to build AI applications in-house, given broad access to foundational models and datasets, rather than buying off-the-shelf solutions. In addition, CTOs have shown reluctance to commit to expensive new products amid a flood of AI software offerings, long evaluation cycles, and rapid innovation that risks rendering first-generation tools obsolete. In short, enterprises appeared to be in “wait-and-see” mode.
However, this dynamic is starting to shift. Most enterprises—particularly SMBs—lack the resources to sustain large-scale custom AI development and will inevitably transition toward commercial platforms as products mature. Evidence of this turning point came through in Snowflake and MongoDB’s results: Snowflake’s product revenue growth accelerated to +32% (from +28% in Q1), while MongoDB’s Atlas database grew +29% (from +26%), with both companies raising full-year guidance. This momentum is consistent with the broader takeoff of AI inference reported by hyperscalers and cloud suppliers, highlighting tangible enterprise use cases. Snowflake in particular noted that customers are investing heavily in data modernization and transformation tied to AI initiatives.
With most data science companies determined to turn into large platforms offering a wide range of storage and data preparation functionalities such as the customization of large language models based on a customer’s proprietary data, new products and use cases should also fuel growth going forward.
Importantly, growth acceleration did not come at the expense of profitability. Snowflake delivered an 11% operating margin (+600bps YoY), while MongoDB posted 14.7% (+370bps YoY), both ahead of expectations. Assuming continued topline momentum and financial discipline, margin leverage should then emerge as an integral part of the investment thesis on these names.
In conclusion, we believe that sentiment for data science software names could materially improve after years of elusive AI upside as their growth trajectories begin to mirror enterprise AI inference. Still, we stick to our view that selectivity is of paramount importance when investing in the AI software space where we see a very few winners and beneficiaries outside of Big Tech companies and many losers.
