{"slug": "twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai", "title": "Twelve Labs raises $100 million as Amazon bets its Trainium chips on video AI", "summary": "Twelve Labs Inc., a video AI startup backed by Nvidia, raised $100 million in Series B funding co-led by New Enterprise Associates and Naver, with Amazon.com participating. AWS signed a multiyear contract to host Twelve Labs' workloads on its Trainium chips, marking a strategic shift away from Nvidia GPUs. The deal underscores Amazon's push to compete with Nvidia in AI infrastructure by securing a fast-growing startup for its custom silicon.", "body_md": "*Twelve Labs just raised $100 million, and the real story isn't the money, it's who's hosting the workloads.*\n\nBloomberg reported on July 1, 2026 that Twelve Labs Inc., the Nvidia-backed startup building AI that searches and understands video, is raising the $100 million Series B round co-led by New Enterprise Associates and Naver, with Amazon.com among the backers. Radical Ventures, Index Ventures and Korea Investment Partners also joined. The San Francisco company was founded in 2021 by Jae Lee and four co-founders, and it has previously drawn backing from Nvidia, Intel and Samsung.\n\nHere's the part that matters most for the infrastructure war underneath. AWS signed a multiyear contract to host Twelve Labs' workloads on its own Trainium chips, and new Twelve Labs models will launch on AWS for developers building video search and analysis tools. That's not just cloud hosting. It's a commitment away from Nvidia GPUs, even for a startup Nvidia itself backed early, alongside Intel and Samsung, in a $10 million round years ago.\n\nLee, who worked as a data scientist for Korea's Ministry of National Defense before starting the company, built Twelve Labs around a gap he says the rest of the AI industry ignored. Everyone was racing to master text and images. Video, the largest and messiest data type of all, sat mostly untouched. Twelve Labs' models let a company search a video library the way you'd search text: find every clip where a product appears, locate the exact moment something happened, flag a safety incident buried in hours of security footage.\n\nAmazon doesn't need to own a video AI company to win here. It needs Twelve Labs running inference on Trainium instead of Nvidia's chips, because every hour of compute that runs on AWS's own silicon is an hour Amazon isn't paying Nvidia's margins. Trainium2 chips already power Anthropic's Project Rainier cluster in Indiana, and locking in a fast-growing startup like Twelve Labs extends that same strategy into a new market.\n\nThat's the tension sitting underneath this deal. Nvidia backed Twelve Labs when it was small and unproven. Now that the startup has raised a serious round, its biggest new customer is the cloud provider building custom chips specifically to compete with Nvidia. Twelve Labs isn't ditching Nvidia hardware everywhere, but the AWS contract shows where the new capacity is actually going.\n\nThe wider pattern matters more than any single funding round. Multimodal foundation model startups have become the proving ground for whoever's chips they choose to scale on. When Amazon can point to a company like Twelve Labs running production workloads on Trainium and shipping models through Amazon Bedrock, that's a stronger sales pitch to the next enterprise customer than any benchmark slide.\n\nNone of it works if the product doesn't hold up. Twelve Labs sells enterprises on a narrow, specific promise: index a video library once, then let anyone in the company query it in plain language instead of scrubbing through footage by hand. Media companies, retailers reviewing security tape, and sports leagues cataloguing highlight reels have all been solving that problem badly with manual tagging for years. If the $100 million buys enough engineering time to make that search fast and accurate at real scale, the Trainium bet pays for itself. If it doesn't, Amazon's chips are just hosting another AI company that burned through a big round without proving the business underneath it.\n\nWhat happens next is easy to watch. Twelve Labs said its new models are rolling out on AWS for developers building video search tools. The real test isn't this funding announcement. It's whether working applications ship on top of it before the year is out.\n\n**Also read:** [Oracle Just Handed Wall Street Its Best Case Against the AI Spending Boom](https://startupfortune.com/oracle-just-handed-wall-street-its-best-case-against-the-ai-spending-boom/) • [Anthropic Gets Its Most Powerful AI Models Back After an 18-Day Fight With Washington](https://startupfortune.com/anthropic-gets-its-most-powerful-ai-models-back-after-an-18-day-fight-with-washington/) • [Tech and Finance Are Cutting Jobs by the Thousands as AI Spending Climbs](https://startupfortune.com/tech-and-finance-are-cutting-jobs-by-the-thousands-as-ai-spending-climbs/)", "url": "https://wpnews.pro/news/twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai", "canonical_source": "https://startupfortune.com/twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai/", "published_at": "2026-07-01 11:44:40+00:00", "updated_at": "2026-07-01 11:53:27.605746+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-startups", "ai-infrastructure", "ai-chips", "computer-vision"], "entities": ["Twelve Labs", "Amazon", "Nvidia", "New Enterprise Associates", "Naver", "AWS", "Trainium", "Jae Lee"], "alternates": {"html": "https://wpnews.pro/news/twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai", "markdown": "https://wpnews.pro/news/twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai.md", "text": "https://wpnews.pro/news/twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai.txt", "jsonld": "https://wpnews.pro/news/twelve-labs-raises-100-million-as-amazon-bets-its-trainium-chips-on-video-ai.jsonld"}}