brain/
← all entities
entitygenericartificial-intelligence

Databricks

Notes

Databricks

One-line summary: Enterprise data platform (lakehouse architecture) + AI model serving / fine-tuning / agentic-deployment vendor. ~6,000-person go-to-market org + 3-4,000-person R&D org as of late 2025. Tracked here because Databricks is the largest operator-grade vendor whose CEO (ali-ghodsi) gives consistent on-record framing on enterprise AI deployment economics — including the "LLM is a commodity" thesis, the three-camps framing, and the December 2025 "we have AGI" baseline view.

What they are

Cloud data platform built on the lakehouse architecture (unified data lake + warehouse). Sells AI infrastructure: model serving, fine-tuning, RAG / vector search, agent runtime, MLflow lineage. Major enterprise customers cited by Ghodsi: Royal Bank of Canada (RBC), Merck, 7-Eleven. Founded by Berkeley AMPLab researchers (ali-ghodsi, Matei Zaharia, Ion Stoica, et al.).

Why they matter to this thread

Databricks isn't a frontier lab and doesn't make consumer AI products. It matters to the thread because:

  1. Operator-side voice on deployment economics. ali-ghodsi's framings (llm-as-commodity-thesis, three-camps-of-ai, "data as moat") are now the wiki's primary anchor for how the enterprise AI deployment layer thinks about value capture. They're a Camp 3 vendor par excellence.
  2. Failed-fine-tuning evidence. Both Databricks and Glean (Arvind's company, same conversation) have first-person evidence of fine-tuning bets that didn't pan out — strong evidence for the commodity thesis.
  3. Internal AI adoption data. Ghodsi gives operator-level adoption signal across GTM, R&D, finance, and HR. Useful proxy for what Databricks-scale enterprises (~10,000 employees) actually deploy.

Evidence

Customer use cases (Dec 2025)

  • ali-ghodsi in 2025-12-23-bg2-databricks-glean-enterprise-ai: "Royal bank of Canada built agents with us that basically take as soon as an earnings report comes out... the agent goes, gets the earnings report, gets all the previous earnings reports, gets all the competitors earnings reports, gets everything that's going on in the market, does the full analysis, the news, everything, puts it all together and it can get the equity report out in 15 minutes from the earnings call. Industry standard is 2 hours." Cited as a working use case in finance.
  • ali-ghodsi in 2025-12-23-bg2-databricks-glean-enterprise-ai (Merck): "Customer Merck that in the life science space created a model called Teddy. Teddy stands for transformer enabled drug discovery... a transformer model, kind of just like large language models that can predict the next word but it instead can figure out which genome is missing if you remove a genome. So it really understands the gene regulatory network." Cited as cutting-edge use case in healthcare.
  • ali-ghodsi in 2025-12-23-bg2-databricks-glean-enterprise-ai (7-Eleven): "711 agents that completely automate the marketing stack... these agents can basically prepare, they can segment the audience like this segment wants to hear this and it can prepare all the marketing material that's like directly targeting you guys and they can put the campaigns together and do that." Marketing-automation as the retail use case.

Internal deployment status (Dec 2025)

  • ali-ghodsi in 2025-12-23-bg2-databricks-glean-enterprise-ai: "Databricks is a big 6,000 person go to market org and 3,4000 person R&D Org and then there's some back office stuff those two already we're seeing heavy automation using agent for all kinds of the tasks... Finance is all on databricks and it's all the forecasting... it's all moved to machine learning based. But it took them a long time because they had their Excel models and they're very proud of them." — Operator-level signal: GTM + back-office finance heavily automated; engineering automation "initially failed not because AI was wrong but because of humans/org structure"; HR + analytical departments lagging.

Strategic positioning

  • Databricks's product evolution aligns with Ghodsi's "value accrues to apps" framing — the company is moving from data infrastructure (commodity-ish) to app-layer products (agent platforms, AI/BI, etc.). Cross-link llm-as-commodity-thesis.

Named on OpenAI ChatGPT Work Data agent (Sep 2026 — issuer list, not a Databricks card)

Sources

Related

Referenced by
Entities
ChatGPT Work Data agentGlean