
LanceDB Banks $57.5 Million Series B for AI Training Data
LanceDB raised $57.5 million led by Bessemer and Innovation Endeavors, bringing total funding to $98.5 million, as revenue grows 10x and downloads pass 120 million.
- By
- Nathan Brooks
- Filed
- Channel
- Startups & Funding
- Read
- 3 min read
LanceDB has raised $57.5 million in a Series B round to manage the training data behind models at Netflix, Microsoft AI, Runway, Luma AI and Midjourney. Bessemer Venture Partners and Innovation Endeavors led the round, with Theory Ventures, Dimension, Swift Ventures, CRV and Y Combinator participating. Total funding now stands at $98.5 million; the company did not disclose a valuation.
The announcement came on Tuesday, Nov. 17. The company builds a data platform for training AI models on images, video and sensor data — the unglamorous plumbing that increasingly decides whether GPU clusters run or idle.
LanceDB said revenue has grown more than 10x year over year since its Series A last June. The company did not give absolute figures, so the multiple is a claim, not a disclosed result. Its open-source library has passed 120 million downloads, roughly six times the total a year earlier.
What does the platform actually do?
The foundation is Lance, an open-source data format that stores structured fields alongside images, video and other large files. It supports fast retrieval and edits to individual records. LanceDB layers a lakehouse on top for search, curation, feature engineering and training, and it reads data in place in customers' existing cloud storage rather than copying it.
Co-founder and chief executive Chang She told SiliconANGLE that research teams lose "weeks or months" assembling a training dataset spread across separate systems. The scale is not trivial: large multimodal training tables commonly hold 10 to 20 petabytes, SiliconANGLE reported, with some approaching 80 to 100 petabytes. LanceDB Enterprise now offers managed compaction that the company says supports tables of up to 100 billion rows, with multiple researchers reading and writing simultaneously.
For perspective, a 100-billion-row table of multimodal data is less a spreadsheet than a small national archive with a throughput problem.
Who uses it, and how deeply?
Microsoft AI runs LanceDB as the data layer for training across billions of rows of image and code data, according to Suqiang (Jack) Song, its head of data platform. Nvidia's Cosmos team, an open-source user and partner, offers another concrete data point. Max Li, technical lead for Nvidia Cosmos, said the team curated roughly a million hours of video with Lance for Cosmos 3 and is now training directly from it.
Netflix, Runway, Luma AI and Midjourney round out the named customer list. All five sit squarely in the generative video and image segment, where training sets are large, messy and expensive to move.
Where does the money go?
LanceDB said the new capital will fund work on data curation, feature engineering and data loading that keeps large GPU clusters fed during training. The roadmap also adds support for video generation, world models and robotics — three workloads where petabyte-scale curation is the norm rather than the exception.
The company is betting that model quality now hinges less on architecture papers and more on dataset hygiene. Its 120 million downloads suggest a fair number of engineers already agree.
The competitive question is whether the incumbents in data infrastructure move fast enough on multimodal formats to make a dedicated lakehouse unnecessary. With revenue multiplying tenfold and Nvidia, Microsoft and Netflix on the reference list, LanceDB has at least bought itself the chance to answer.
Source: mlq.ai
More from Nathan Brooks
Show full bio
Senior reporter covering people, launches and day-to-day practice in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,056 articles
Related articles
mistral-ceo-targets-european-ipo-rules-out-listing
Mistral CEO Targets European IPO, Rules Out 2025 Listing
pedestal-ai-exits-stealth-with-4-7m-seed-to-put-ai-agents-on-delivery-routes
Pedestal AI Exits Stealth With $4.7M Seed to Put AI Agents on Delivery Routes
model-ml-raises-250m-at-1-5b-to-automate-junior-banker-work
Model ML Raises $250M at $1.5B to Automate Junior Banker Work
cramer-warns-anthropic-ipo-riches-flow-only-to-the-super-rich
Cramer Warns Anthropic IPO Riches Flow Only to the Super Rich



