Reads end 15 days later, on 22 December. Open-source ChromaDB stays under Apache 2.0, and code from private repositories has been added to it, but hosted customers have to export and move.
LanceDB raises $57.5 million Series B for its multimodal lakehouse
EditorialBy TrustList Editorial
LanceDB says it has raised $57.5M in Series B funding to bring its multimodal lakehouse to researchers, model builders and platform teams; press reports name the investors and a $98.5M total.
About LanceDB raises $57.5 million Series B for its multimodal lakehouse
LanceDB raises $57.5 million Series B for its multimodal lakehouse
17 November 2026: LanceDB, the company behind a data platform for AI research and model training, has announced a $57.5 million Series B. Its own blog banner reads that AI research has outgrown the modern data stack and that it raised the money to rebuild it, and its post list describes the funds as bringing its multimodal lakehouse to researchers, model builders and platform teams working at the frontier of AI.
Not yet independently verified. LanceDB's own page shows the $57.5M Series B headline and little more, with no date printed. The lead investors, the other investors, the $98.5 million total, the date of 17 November 2026 and the chief executive's quote all come from one press report (SiliconANGLE) and were not confirmed with the company. We did not confirm where the company is based. We will update this when it can be confirmed, and remove this note.
What the company page confirms
The company's blog carries the headline and a link to a longer story, but the page we could read gives no investor names, no total raised and no breakdown of how the money will be spent. Everything below beyond the headline figure comes from press coverage, and the note on this page says so. The same blog list shows a previous round described as a $30 million Series A, which tells a buyer how quickly the company has scaled its funding.
Investors and total funding, as reported
SiliconANGLE reports that the round was led by Bessemer Venture Partners and Innovation Endeavors, with Theory Ventures, Dimension, Swift Ventures, CRV and Y Combinator also taking part. It puts the company's total funding at $98.5 million. The same report quotes co-founder and chief executive Chang She saying there is a missing data layer for AI today, and that this is what the company is trying to build.
The product behind the round
According to the press report, LanceDB is built on Lance, an open-source data format designed for AI workloads. The company's own blog posts, listed on the page we read, give a sense of the use cases it promotes: robotics training that reads Lance tables straight from object storage, distributed vector search across ten billion vectors at more than a thousand queries per second, pretraining a language model from one table, and an enterprise product used by an investigations platform for semantic search and ranking. Those are the company's claims in its own posts, and we have not benchmarked any of them.
What a data or platform team should do
A funding announcement is not a reason to switch tools, but it is a reason to look again if you store training data, embeddings or sensor data in several systems at once. Start by writing down what your team keeps today: raw files in object storage, tables for metadata, a vector index for search, and the glue between them. Then ask where those layers disagree, for example when an embedding is out of date or a dataset version cannot be reproduced.
If that describes your pipeline, a multimodal lakehouse is one of several designs worth testing next to your current one. Run a small evaluation on your own data rather than relying on a vendor's figures, and measure what matters to you: load time into training, query latency at your real scale, cost of storage, and how easily you can roll back to an earlier version of a dataset.
Because the format is open source, a buyer can also check how much of the stack they could run themselves, and what is reserved for the paid enterprise product. Ask the company in writing about licence terms, support commitments, security certifications, data residency and the exit path if you later want to move your data out.
Finally, treat the investor list and the total funding as reported figures until the company's full announcement is available to read. If you plan to depend on a vendor for several years, its financial position is one input, but its customer references and the openness of its format tell you more about the risk of lock-in.
Related on TrustList:
- Chroma Cloud goes read-only on 7 December as Chroma joins Replit
- DuckDB 1.4 support ends 17 November as 2.0 LTS ships
- Gemini API retires its embedding preview and reroutes the calls
Sources
Categories & features
TrustList Weekly
The week in software and IT, in one email
The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.
You may be interested in
DuckDB 2.0 is supported until at least October 2027 and 1.5 has been extended to March 2027. Files written by 2.0 cannot be read by any older version, so upgrade order matters.
Google says the replacement keeps the same schema, pricing and embedding space, so no re-indexing is needed. Calls to the preview string now bill at Gemini Embedding 2 rates.
Most of the roles abused were old integrations nobody was using any more. Wiz advises searching CloudTrail from 1 October and rotating every secret those roles could read.
New jobs compute starts in Auto access mode, a Genie One MCP endpoint stops answering on 30 November, and SQL AI Functions need Runtime 17.3 LTS from 1 February 2027.
Arity, the mobility data company of Allstate, is buying substantially all of INRIX's assets and expects to close in 2026. No price has been disclosed.
CREATE OR REPLACE VIEW keeps grants from 2 December. Databricks has since dropped its 1 November date for removing the partner AI toggle and postponed the OpenSharing URL change with no new date.
Five categories hold half of everything on the platform. Here's what the real shape of the catalogue looks like - and what it means for how to search it.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- More Vector DatabaseThe ranking for this subject
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.