Snowflake, Databricks, and the Cloud Data Market
Bonus: SaaS Disrupted?
SaaS Disrupted?
We’ve been big software bulls over the past few decades, but the rise of AI and LLMs will likely have a profound impact on the industry. At this stage it’s not clear yet to what extent positives will outweigh negatives. Having recently done a large coding project, working together with LLMs probably boosted productivity with around 4 to 5x. Thus, an obvious bear case would be that increasing productivity levels in the coming decade will lead to a reduction of software seats that enterprises will have to purchase. Another bear case is that SaaS apps have an abundance of functionality built-in through massive logic-based code bases in the back, however, if users start interacting more with LLMs via natural language, a lot of this code might actually become obsolete. Basically, the business application of the future could be more like a portal where a user interacts with an AI agent, which carries out all tasks while interacting with a database in the back.
This is Satya Nadella, Microsoft’s CEO, on the potential disruption of SaaS on the B2G podcast:
“Business applications are essentially CRUD databases with a bunch of business logic. The business logic is all going to these agents and they’re going to update multiple databases, and all the logic will be in the AI tier so to speak.”
In a subsequent interview with The Economic Times, Nadella stated:
“A lot of the business logic will move to a new tier, which then will be a multi-agent tier that needs to be orchestrated. It's going to be an agent that will orchestrate across multiple SaaS applications. Just like I can build a spreadsheet, I will build thousands, hundreds of agents that are working to help me streamline my own work”
However, it’s not all doom and gloom in the outlook for SaaS. Strong controls will have to be implemented, likely via business logic, for example, ‘this user is only allowed to interact with this part of the database’ etc. It’s also likely that common workflows where a human needs to be involved can easier be done via pre-built business logic. For example, even though the AI will be writing a lot of code, we still need an IDE because humans need to validate and adapt the code. Similarly, we still need software to manage and deploy the software, which is largely running on traditional logic. In the same vein, we still need Microsoft Office. Even though AI might be preparing a lot of Powerpoints and Excel sheets, humans again still need to review and adapt them. Thus, the future we see in the coming five years is one where AI will be increasing automation in a wide variety of fields, but traditional software will still be needed to manage overall workflows.
Another positive is that lost user seats from higher levels of automation can be taken up by companies in new industries that will emerge, which has been the historical trend. A few hundred years ago, 90%+ of the population was employed in farming. Large gains in agricultural productivity, and subsequently industrial productivity, didn’t lead to mass unemployment. Instead, a vast services industry emerged where an abundance of skilled workers were needed ranging from finance to tech and health care. Similarly, software apps are also introducing automated workflows within their environments thanks to AI and this compute time can start being monetized over time.
Another positive from AI for the SaaS names is that their skilled workforces are highly expensive, resulting in massive salaries and share based compensation (SBC). If worker productivity can be improved with 2x or more over time, there’s a lot of potential for SaaS companies to become leaner organizations and boost earnings margins. Not only can coding productivity see a massive boost, but also services productivity as more customer queries can get handled by an AI. Similarly, for new software implementations, an AI teacher can get new users up and running in much faster ways, and the AI can also write customized software extensions which enterprises often need.
When it comes to software investing, historically, most of the outperformance has been driven by strong top line growth rates. However, there is a good reason to believe that with AI, investors can also start being much more enthusiastic when it comes to margins.
The Bull Case for the Cloud Data Market
In the remainder of this article, we will zoom in on the cloud data market for analytics. The best long term bull case for Snowflake, and its peers such as Databricks, was probably given by Intel’s former CEO a few years ago, who was running virtualization software company VMware previously:
“We are in year 20 of the public cloud, and you have 60%-plus of compute in the cloud, but 80%-plus of data remains on-prem.”
For premium subscribers, we will discuss Snowflake’s current battle with Databricks for the cloud data market, and the company’s future opportunities. We will also give the detailed insights from a large number of well known, blue chip enterprises on how they’re using cloud data systems and where they are increasingly moving workloads to. And we’ll discuss the same topics with a VP at Accenture, one of the largest system integrators in the world. Finally, we’ll do an analysis of the financials and valuations for Snowflake as well as names in the wider software space, and we’ll conclude with our stock picks.
Snowflake vs Databricks
Although historically both Snowflake and Databricks came from different backgrounds—Databricks was oriented towards running data science workloads in the cloud while Snowflake was largely a horizontally scalable SQL-based data warehouse—both companies have been moving into each other’s territories over the last years. Snowflake has been introducing support for unstructured and semi-structured data, in order to transform its cloud platform to become more of a data lakehouse like Databricks. Snowflake introduced also Snowpark so that data engineers can run data workloads on Spark-like clusters via a variety of popular data science languages such as Python, Java and Scala; which is really the bread-and-butter of Databricks. At the same time, Databricks has been introducing support for SQL and ACID transactions to allow its data lake to also function as a cloud-based data warehouse, naming it ‘the data lakehouse’. Traditional business analytics built on top of SQL and ACID transactions remains the core of Snowflake’s business today.