Over the last few years, the heat in the data market has been attracting hundreds of new companies, each looking to cut their slice out of a ballooning pie (View Highlight)
Note: Like this visual of ballooning pie
In our current ecosystem, most data products are still expensive to build. They require architecting new frameworks, developing smarter AIs, designing complex visualization systems, or reinventing how data gets processed. Work like this isn’t cheap, and companies can only fund it if they promise to make real money on the other side. This sets a high floor for the price of data products (View Highlight)
Furthermore, while high prices are a constant, data companies’ business models are not. Some products charge usage-based compute fees, some for user licenses, some for feature tiers, and some blend different mechanics. Some products are still searching for the best way to sell. And some, which bury their prices behind a handful of sales calls, don’t tell us what they charge for. (View Highlight)
The more common comparison is that data products (and data teams) are tracing the same path as web development stacks, which are also interconnected mobiles of tools delicately balanced against one another. In time, it’s implied, our stacks will find the same equilibrium.
But the analogy is fatally flawed. Software stacks are built by engineers, who are comfortable wiring together closer-to-the-metal technologies. Many data practitioners are unwilling or unable to do the same. They want to buy hosted services. If we were generally comfortable managing Spark ourselves, Databricks wouldn’t be worth nearly $40 billion (View Highlight)
Second, open-source standards have a harder time taking hold. While LookML could’ve become a standard for semantic modeling, Looker had little incentive to open it up because most people wouldn’t want or be able to run it themselves. An open-source LookML is a gift to Looker’s competitors, not to the community (View Highlight)
These aren’t the inevitable poles, though. There are potentially other, less severe ways to balance the economic equation.
Middle men could insert themselves between today’s vendors and buyers, offering everything from shopping catalogs of data tools to a layer of software to manage those tools. A number of consultancies already do this manually; the next wave of companies will do it automatically. (View Highlight)
Note: Can you curate a data stack for others?
there could be a rise in soft M&A, in which companies white-label themselves behind other vendors and through revenue sharing agreements. These arrangements would probably be beneficial to the market, but are surely a nightmare to work out (View Highlight)
Companies offer platforms for other people to build on; this enriches the market by making it cheaper for everyone to build new products; those products simultaneously fill the platform’s gaps and reinforce the platform’s dominance. (View Highlight)