Twelve Labs, a US developer of AI specialized in video understanding, is raising $100 million in a Series B round co-led by NEA Management and Naver Ventures. Amazon.com is participating as an investor, and a multiyear contract with AWS will host the company's workloads on Trainium chips.
July 1, 2026 · Twelve Labs · Series B
$100M to make the world's video as searchable as text
Video-AI startup Twelve Labs raises a Series B co-led by NEA and Naver Ventures — with Amazon (AWS) signing a multi-year deal to host its workloads on Trainium chips and offer its models on Bedrock.
$100M
Series B round, co-led by NEA and Naver Ventures
~80%
of the world's data is video — mostly unsearched
47
languages supported by the Marengo model
Indexing speed
One hour of video indexed in about a minute — roughly 60× real time
Capacity to process more than 10,000 hours of video per day .
Marengo 2.7
Generates embeddings integrating visual, audio and text — powering natural-language search and similarity detection.
Pegasus 1.5
Reasons continuously over a video's timeline for summarization, metadata generation and story understanding.
Building the funding base
Jun 2024
$50M · Series A
Co-led by NEA and NVIDIA NVentures, plus Index Ventures
Dec 2024
$30M · Strategic
Databricks, Snowflake, SK Telecom, HubSpot Ventures, IQT — cumulative funding passes $100M
Jul 2026
$100M · Series B
Co-led by NEA and Naver Ventures, Radical Ventures, Index Ventures, Korea Investment Partners
What's working
Manual search cut from days to seconds
Brand-safe contextual targeting
Easier semantic search and RAG on large video datasets
Open challenges
Context accuracy for long videos
Need for domain-specific fine-tuning
Competition from video features in general LLMs
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…