The Data Cleanup Trade
Reva Jariwala@reva_jariwala
−3.3%
since Oct 2
The idea
I believe the real bottleneck in enterprise AI isn't the chatbots, it's the years of messy data behind them, so the money flows to whoever cleans it up. This basket owns all three layers: the platforms companies consolidate data onto, the integrators paid to do the dirty work, and the plumbing plus storage underneath it all. It leans on large, steady anchors with a few sharper smaller bets around them, and keeps 10% cash to add if enterprise software gets cheap. This is built as a multi-year grind, not a quick catalyst.
What it owns
- The places companies consolidate their messy data to make it usable for AI. · SnowflakeIt is where clients consolidate scattered data into one place before building AI applications, so growth depends on steady product revenue from large enterprise accounts., MicrosoftAzure and Fabric form the default data stack for companies shipping AI work, so the take requires sustained growth in data and AI business spending separate from chatbot adoption., OracleIt already holds decades of enterprise data that gets modernized for AI, so the take needs cloud database and infrastructure revenue to climb as companies migrate old systems., MongoDBIt is the database developers choose when rebuilding applications around cleaner data, so the take works if Atlas adoption stays healthy on new AI-adjacent projects., AlphabetBigQuery is one of the main warehouses where companies consolidate data before AI work begins, so the take needs Google Cloud to keep winning enterprise data workloads.34%
- The firms companies actually pay to untangle their data and get AI projects shipped. · PalantirIt embeds at clients to untangle messy data into usable form, so the take holds if commercial customer counts and deal sizes keep expanding., AccentureIt provides services when companies cannot clean their own data internally, so the take needs AI-related bookings to keep rising as spending flows to the cleanup work., IBMIt earns from both consulting services and software platforms that run cleanup work, so the take requires sustained growth in consulting AI bookings and software revenue., InfosysIt handles the labor-intensive data migration and restructuring work at offshore cost levels, so the take needs enterprises to keep outsourcing cleanup rather than cutting IT budgets., EPAM SystemsIt specializes in outsourced data engineering work, so the take holds if demand for skilled pipeline builders keeps recovering as shown by rising revenue.28%
- The pipelines and monitoring tools that move the data and catch it when it breaks. · DatadogIt provides the monitoring layer companies add once data starts moving and AI systems expand, so the take needs customers to keep adopting more products., ElasticIt makes cleaned-up data searchable and sits under AI retrieval systems, so the take requires sustained spending as companies wire up AI features over their own data., DynatraceIt monitors whether data pipelines and AI systems work correctly in production, so the take requires large enterprises to keep expanding observability spending as complexity grows., InnodataIt prepares messy data specifically for AI models, so the take only works if large customers keep signing data-prep contracts and driving revenue growth.17%
- All that cleaned-up data has to physically live somewhere. · Dell TechnologiesIt sells the servers and storage infrastructure where consolidated data sits, so the take requires enterprise hardware refreshes to continue for AI workloads., NetAppIt provides the storage layer for data spread across on-premises machines and cloud environments during cleanup, so the take needs hybrid storage demand to remain firm.11%
- Cash10%