On July 1, 2026, video-understanding AI startup TwelveLabs announced it closed a $100M Series B. On the surface that's just another AI funding headline. But look at who co-led the round: NEA, the storied US venture firm, standing shoulder to shoulder with NAVER Ventures. That's right — a Korean tech giant's VC arm didn't just write a check, it co-led a US-based frontier AI round in one of the hottest categories on the planet right now: teaching machines to actually understand video.

And it doesn't stop there. Korea Investment Partners (KIP) also joined as a participant. So picture the cap table: Amazon, Radical Ventures, Index Ventures, Quadrille Capital, and Red Bull Ventures all sitting alongside NAVER and KIP. Why does this particular round matter? Because a startup that looks, on paper, like a classic Silicon Valley story turns out to have deep Korean roots — a sizable Seoul office, founders with strong Korean ties, and now a capital structure that makes that connection official.

If "video AI" sounds vague, here's the plain-English version. AI has gotten pretty good at reading text and decent at looking at still images, but video has remained the hard problem. Finding one specific incident buried in hours of security footage. Pulling the exact highlight clip out of a full soccer match. Sorting through tens of thousands of ad creatives to find the ones where your logo actually shows up on screen clearly. That's the kind of grinding, manual work TwelveLabs wants to automate — and with this raise, the company is explicitly saying it wants to go beyond "video understanding" and build something closer to video superintelligence.

The Players

The company at the center of this is TwelveLabs, and the way it's framing its own ambition this time is worth noting. It's not calling itself a video search tool anymore. It's positioning itself as building a full-stack "agentic intelligence system for video" — one that fuses perception, knowledge, and reasoning into a single architecture. Strip away the jargon and what they're really saying is: an AI that doesn't just watch a video, but understands the context inside it well enough to reason about what it's seeing.

Co-leading the round are NEA and NAVER Ventures. NEA (New Enterprise Associates) is one of the largest, longest-running venture firms in the US, with a track record spanning everything from autonomous vehicles to biotech. NAVER Ventures is exactly what it sounds like — the venture arm of NAVER, Korea's biggest search-and-platform company. The headline fact here is that a major Korean tech company put its own capital behind co-leading a Series B for a US video AI startup. That's not a small gesture.

The participant list reads like a who's-who. Amazon jumping in signals that a company touching video across cloud infrastructure, e-commerce, and content is taking TwelveLabs' technology seriously enough to back it directly. Radical Ventures brings AI-specialist credibility, while Index Ventures and Quadrille Capital are established global investors with broad portfolios. Then there's a name that stands out for a different reason: Red Bull Ventures, the investment arm of a company steeped in sports and entertainment content — which, as you'll see later, lines up neatly with one of TwelveLabs' target industries.

On the Korean side, Korea Investment Partners (KIP) rounds out the picture as a participant. KIP is one of Korea's most established VCs with a long track record in the local startup ecosystem. So the shape of this deal is: NAVER Ventures takes a lead position, KIP adds weight as a participant, and underneath all of it sits the fact that TwelveLabs itself has real Korean roots — a founding team with strong Korean ties and an already-substantial Seoul presence that made this kind of investment feel natural rather than opportunistic.

What Happened

Let's lay out the facts as announced. On July 1, 2026, TwelveLabs closed a $100M Series B, co-led by NEA and NAVER Ventures. Participants included Amazon, Radical Ventures, Korea Investment Partners, Index Ventures, Quadrille Capital, and Red Bull Ventures. Worth noting: no specific valuation was disclosed in this announcement.

TwelveLabs was clear about where the money is going. At the core is advancing two flagship models. The first is Marengo 3.0, a video embedding model. The second is Pegasus 1.5, which converts raw video into structured data — scene boundaries, entities, temporal segments, and semantic context. The company bundles these together under a broader concept it calls the "Video Cognition System."

Here's a quick-reference table for the round.

Item Detail
Announced July 1, 2026
Round Series B, $100M
Co-leads NEA, NAVER Ventures
Key participants Amazon, Radical Ventures, Korea Investment Partners, Index Ventures, Quadrille Capital, Red Bull Ventures
Core models Marengo 3.0 (video embedding), Pegasus 1.5 (structured data conversion)
Expansion Scaling San Francisco and Seoul; new offices in New York and London
Target verticals Advertising, security, sports, automotive

The use of funds goes beyond model development. TwelveLabs also wants to scale its Video Cognition System into what it calls the world's most important video archives, and it's investing in team growth and R&D along the way. Geographically, the plan is to grow its existing San Francisco and Seoul teams while opening brand-new offices in New York and London. Put those four cities together and you get a genuinely global footprint spanning the US West Coast, Asia, the US East Coast, and Europe.

The target industries are worth sitting with for a second: advertising, security, sports, and automotive. What do these four have in common? Each one generates massive volumes of video and has historically relied on humans to manually sort through it. Advertisers need to find which creative patterns actually perform across tens of thousands of assets. Security teams need to locate specific incidents inside hundreds of hours of footage. Sports needs highlight reels pulled out of full-length game footage. Automotive needs to analyze driving footage at scale. All four are exactly the kind of real, monetizable demand TwelveLabs is chasing.

What Each Side Gains

Start with TwelveLabs itself. This round is worth more than the dollar figure. Having a top-tier US VC like NEA and a major Asian platform company like NAVER co-lead at the same time signals that the company has real footing on both sides of the Pacific — not just Silicon Valley credibility, but Asian market credibility too. And with Amazon in as an investor, the door to future infrastructure or product partnerships stays wide open. In effect, TwelveLabs walked away with several stamps of validation at once, not just a wire transfer.

NAVER's angle is really the heart of this story. NAVER is a sprawling Korean platform business touching search, commerce, content, and cloud — and it's sitting on an enormous, ever-growing pile of video assets across short-form content, live commerce streams, and news footage. Building the technology to truly understand and search through all that video in-house would take significant time and money. By having NAVER Ventures co-lead this round, NAVER gets an early, close-up view of TwelveLabs' technology and effectively positions itself to be first in line if that tech gets woven into NAVER's own products down the road. And because TwelveLabs already has Korean roots and a growing Seoul team, the collaboration barrier is lower than it would be with a fully foreign, unconnected startup.

For Korea Investment Partners, this is an attractive bet on its own merits. Getting into a global frontier AI startup at the relatively de-risked Series B stage, alongside names like NEA and Amazon, isn't an opportunity that comes around often for a Korean VC. It opens up one more channel for domestic capital to ride alongside top-tier global deep tech deals. As for Amazon, the logic is straightforward: across AWS, e-commerce, and content businesses like Prime Video, Amazon needs video-understanding technology in multiple places — so this is as much a strategic technology bet as it is a financial one.

Finally, there's real upside for the advertising, security, sports, and automotive industries themselves. These sectors have poured enormous amounts of manual labor into tagging and searching video content. As TwelveLabs' technology matures, much of that grunt work could shift to AI, cutting both cost and turnaround time significantly. Red Bull Ventures' presence on the cap table is a small but telling signal — a company deeply embedded in sports and entertainment content appears to already be validating this technology's real-world potential, not just its theoretical promise.

Precedents: Wins and Failures

Video AI as a category has matured much later than text or image AI, and there's a structural reason for that: video adds a time axis that images simply don't have. A still image is a single static slice of information, but video is a continuous stream of millions of frames changing over time — which means the sheer volume of data to process is dramatically higher. For years, "video search" mostly meant relying on captions or metadata rather than the actual visual content, and AI that genuinely understands what's happening inside a video has only become practically useful in the last few years.

On the win side, platforms like YouTube and TikTok have applied video-understanding technology to content recommendation and copyright filtering with enormous efficiency gains. These companies built massive in-house video AI infrastructure that analyzes and classifies uploaded content in near real time. You can think of that as a generalized, closed-loop version of what TwelveLabs is calling its Video Cognition System. The key difference is that those systems operate inside a single closed platform, while TwelveLabs is trying to open this capability up as an API or platform that any industry can plug into.

On the failure side, plenty of AI startups have stumbled by promising a sweeping vision that outran what the underlying technology could actually deliver. In multimodal AI especially, the gap between "theoretically possible" and "accurate enough that companies will actually pay for it in production" has repeatedly turned out to be wider than founders expected. Flashy demos that don't hold up under real-world usage have sunk more than a few well-funded startups. TwelveLabs naming specific model versions and technical specs — Marengo 3.0, Pegasus 1.5 — in this announcement reads like a deliberate move to avoid exactly that trap: vision without a shippable product.

From the angle of Korean corporate investment into overseas deep tech, there's already a pattern of Korean conglomerates — NAVER included — making strategic bets on Silicon Valley AI startups, often trying to get early access to technology they can eventually bring in-house. What makes this deal stand out is the weight of the position: not a small follow-on check, but a co-lead role in a company that already has genuine Korean roots baked into its founding story. That combination is relatively rare.

Rivals' Counterplay

The video AI space already has heavyweight players moving fast. Google, Meta, and OpenAI are all continuously expanding video-understanding capabilities inside their broader multimodal models, and their advantage is obvious — massive compute and massive proprietary data. For a specialist like TwelveLabs, going head-to-head with these giants directly isn't really an option. The strategy instead has to be depth: out-specialize the generalists by being dramatically better at video specifically. Emphasizing dedicated model lines like Marengo and Pegasus in this funding announcement fits squarely into that playbook.

The big platforms' counterplay is predictable. They'll keep folding video-understanding capability into their general-purpose multimodal models — think Gemini, GPT-family models, and similar — in an attempt to absorb the reason specialized startups like TwelveLabs exist in the first place. That trend is already underway, and it means specialist companies have to keep proving that a domain-specific model beats a general-purpose one on accuracy and speed within its niche. TwelveLabs zeroing in on advertising, security, sports, and automotive as concrete verticals looks like a deliberate bet on winning through domain-specific precision that general models will struggle to match.

Inside Korea, the competitive backdrop matters too. Kakao and other domestic platform companies are also building out their own AI capabilities. NAVER choosing to invest in TwelveLabs rather than building everything from scratch suggests a calculated bet: that partnering with an already-proven external player is faster and more efficient than reinventing this technology internally. It's a signal that the "build in-house vs. invest and partner" split will keep coexisting as competing strategies among Korea's big tech players.

From a global VC lens, this round is a notable signal too. When a strategic corporate investor like Amazon and a mix of purely financial VCs all pile into the same round, it tends to mean conviction in the category runs wide, not just deep in one investor's thesis. For competing startups, the bar has quietly moved. It's no longer enough to claim "we do specialized video AI too" — you increasingly need TwelveLabs-level specificity: named model versions, clearly defined target industries, and a roster of top-tier backers, or you risk losing the fundraising race entirely.

So What Changes

If you're working in Korea's AI industry, this is a useful case study in how Korean capital rides along with global frontier AI. NAVER Ventures taking a co-lead position suggests this is likely to evolve into more than a purely financial relationship — there's real room for a technology partnership to follow. And a growing Seoul team means actual hiring and collaboration opportunities are opening up domestically, not just abroad. If you're weighing a career in video AI or multimodal AI, TwelveLabs' Seoul office just became a name worth watching closely.

If you're a developer or company plugged into NAVER's ecosystem, it's worth keeping an eye on whether TwelveLabs' technology eventually gets woven into NAVER's search, shopping, or content products. Nothing concrete has been announced on product integration yet, but given how close a co-lead investment relationship tends to be, it's a reasonable next step to anticipate. Think live commerce, short-form video, and news footage search all getting meaningfully more sophisticated at actually understanding what's happening inside the video, not just matching text metadata around it.

If you work in advertising, security, sports, or automotive, treat this as an early signal that how your industry handles video data could shift within the next few years. Work that today involves people manually tagging and searching through footage is trending toward AI automatically structuring scenes, entities, and context for you. Nothing changes tomorrow, but as this technology matures, expect real workflow disruption in how video gets processed across these industries.

If you're on the investing or VC side, this round fits into a broader pattern worth watching: Korean VCs landing co-lead positions alongside top-tier US firms on the same deal, rather than just tagging along on later follow-on rounds. Historically, Korean capital tended to show up after a company was already de-risked. Now you're seeing Korean firms secure co-lead status at the Series B stage itself. Whether this is a one-off or the start of a real trend will take a few more deals to confirm — but at minimum, this is one meaningful data point worth filing away.

🥄 Three Things You're Probably Wondering

— What's TwelveLabs actually worth after this round? The announcement didn't disclose a specific valuation. All that's confirmed is the $100M Series B size — the exact company valuation wasn't released, so it's too early to say.

— Is NAVER going to plug TwelveLabs' tech straight into its own products? No concrete product integration plan has been announced yet. Given how close a co-lead investment relationship tends to be, collaboration seems plausible, but whether and when it actually happens remains to be seen.

— When do the New York and London offices actually open? The announcement confirmed the plan to open new offices but didn't give a specific timeline. Keep an eye on job postings or follow-up announcements for the actual dates.

References

Numbers and criteria are as of announcement and may change. Investment calls are yours to make!