AIToday
Large Language ModelsAI Business & IndustrySiliconANGLE AIPublished: Nov 18, 2026, 04:00 JST

LanceDB lands $57.5 million to train AI on multimodal data in place

Send on LINE
LanceDB lands $57.5 million to train AI on multimodal data in place

3 Key Points

  1. What happened

    LanceDB raised $57.5 million in a Series B led by Bessemer Venture Partners and Innovation Endeavors, bringing total funding to $98.5 million. Its platform lets researchers access, prepare and train on image, video and sensor data without moving it.

  2. Why it matters

    Researchers can spend weeks or months assembling and moving training datasets. Keeping data in place cuts that preparatory work, which may speed up how quickly models reach higher quality.

WHO IT HITSAI research and platform teams working with multimodal datasets at petabyte scale—especially those training video, robotics or world models—gain a single access layer instead of juggling separate storage systems.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

LanceDB's pitch rests on a specific technical choice: an open-source data format called Lance, which is built to handle individual records rather than whole tables. That makes it suited to searching and preparing massive training datasets, where a single table can run 10 to 20 petabytes, and some approach 80 to 100 petabytes. The company has wrapped that format in a lakehouse — a system that combines the storage capacity of a data lake with database-style querying — and added multimodality so text, images, video and sensor data sit in one place.

The practical problem it targets is preparation time. Research teams typically scatter parts of a dataset across systems and spend weeks or months getting a training set ready for a single run. LanceDB's approach keeps data in customers' existing cloud storage and distributes computation to it, so nothing is copied or converted. Runway, Netflix and Microsoft are named as customers already using the platform at scale.

With the new funding, LanceDB says it will focus on large-scale data curation, feature engineering and data loading to keep thousands of GPUs supplied during training. It also plans to deepen support for video generation, world models and robotics, workloads where first-person sensor data and synthetic data add volume and complexity.

FAQ
Who led LanceDB's latest funding round?
Bessemer Venture Partners and Innovation Endeavors co-led the Series B round. Other investors included Theory Ventures, Dimension, Swift Ventures, CRV and Y Combinator.
How does LanceDB's platform handle data differently?
It connects directly to customers' existing cloud storage, and the data stays where it is — no copying or moving. Computational workloads are distributed across the infrastructure needed to process the data.
Which customers already use LanceDB?
Runway AI uses its enterprise platform for AI training, Netflix built a media data lake on the technology, and Microsoft uses it as a data layer for training models on tens of billions of rows of image and code data.
SiliconANGLE AIRead Original Article
Send on LINE

Also reported by Tomasz Tunguz (Theory Ventures)

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleCheng-wen Wu: Jensen Huang's Nvidia return unclear, standards in doubt