Twelve Labs, a startup focused on AI systems for understanding and searching video, has raised $100 million in new funding, according to reports from Bloomberg and PYMNTS.com. Bloomberg reported that Amazon joined the round alongside venture investors, underscoring rising investor interest in infrastructure that can turn large video libraries into searchable, machine-readable data.

The financing matters because video remains one of the hardest data types for AI systems to parse reliably at enterprise scale. Text and images have become standard inputs for modern models, but long-form video introduces cost, latency, and accuracy challenges around scene changes, audio, context, and temporal reasoning. A large round for Twelve Labs suggests investors see a meaningful market in tools that can index, retrieve, and analyze video for applications beyond consumer media search.

Why investors are leaning into video understanding

The reported raise arrives as enterprises accumulate more video than most teams can practically review by hand. That includes marketing footage, customer support recordings, training libraries, security feeds, internal meetings, and entertainment archives. For builders, the commercial question is straightforward: if AI can make video searchable with useful precision, it becomes easier to build products for discovery, moderation, compliance, ad targeting, asset management, and workflow automation.

That is the gap Twelve Labs is trying to fill. While the source materials available here do not provide a detailed product announcement, both reports point to the company’s core positioning around video search and analysis. In practical terms, that puts Twelve Labs in the part of the AI stack that turns raw video into structured signals applications can query.

Bloomberg’s framing of the company as an “AI Video Search Startup” is notable. It suggests investors are not only funding model development, but also the surrounding retrieval layer needed to make video useful in production systems. For many enterprise buyers, search is the first monetizable use case because it solves a direct productivity problem without requiring fully autonomous generation or editing.

Amazon’s participation also stands out. The Bloomberg report says the round included Amazon and VC funds, though the source extract provided here does not specify which Amazon entity invested or whether the investment has any direct commercial tie-in to Amazon Web Services. Without those details, it would be premature to infer a product partnership. Still, strategic interest from a company with deep cloud, media, and AI businesses will draw attention across enterprise AI and developer infrastructure markets.

What video AI has to solve in the real world

Video understanding is attractive on paper, but difficult in deployment. A system has to capture not just objects in frames, but actions over time, spoken dialogue, background sounds, scene transitions, and the relationship between those elements. It also has to do that cheaply enough for customers with large archives and reliably enough that users trust the results.

That is why startups like Twelve Labs are being watched closely by teams building media tools and internal enterprise systems. A video index that misses important moments or returns vague results is much less useful than a text search engine. For product teams, the challenge is not only model quality but end-to-end usability: ingest pipelines, retrieval speed, metadata quality, permissions, and APIs that developers can integrate into existing applications.

The opportunity extends beyond media companies. In enterprise AI, video is often a trapped asset. Businesses may have thousands of hours of recordings but no easy way to find the one product demo, training clip, support interaction, or safety incident they need. If a platform can make those archives searchable and analyzable, it can support workflows in compliance, operations, support, and knowledge management.

That helps explain why a large financing round for a company in this category lands at a moment when AI buyers are shifting from experimentation toward measurable workflow value. Search and retrieval are easier to justify than many open-ended generative deployments because the return on investment can often be framed in saved labor, faster response times, or better asset reuse.

The funding signal and the competitive backdrop

The reported $100 million round is significant even without a fuller public breakdown of valuation or investor mix in the source extracts. It places Twelve Labs among the better-capitalized startups pursuing multimodal infrastructure, a category that spans model providers, vector database vendors, media tooling companies, and application-layer developers.

Competition in this area is not limited to dedicated video startups. Large model providers are steadily improving multimodal capabilities, which means video analysis could increasingly become a feature inside broader AI platforms rather than a stand-alone market. That creates a strategic question for Twelve Labs and similar companies: whether to compete on specialized accuracy and tooling, or risk being subsumed by general-purpose platforms.

That broader platform pressure includes cloud vendors and model companies that are investing heavily in multimodal AI. Amazon, as reported by Bloomberg, is now directly relevant to the story as an investor. Amazon Web Services already serves many enterprises with AI and media infrastructure, so any startup it backs in this space will be scrutinized for signs of ecosystem alignment, even if none have been publicly confirmed in the source materials here.

For founders, the round also signals that investors still see room for focused infrastructure companies in AI, provided they target a hard enough technical problem and a clear enterprise workflow. The market has become more skeptical of thin wrappers around foundation models, but less skeptical of systems that address complex data types and operational bottlenecks.

Evidence, claims, and what is still unclear

The confirmed facts available from this source cluster are limited but consistent across both reports: Twelve Labs raised $100 million, and Bloomberg reported that Amazon participated along with VC funds. PYMNTS.com separately reported that Twelve Labs raised $100 million to fund its bet on video AI.

Several important details are not present in the source extracts provided here. There is no disclosed valuation, no full investor list, and no official statement in the evidence set describing how the capital will be used beyond the broad implication of expanding the company’s video AI efforts. There are also no new benchmark results, customer counts, revenue figures, or product launch details in the available material.

That means readers should be careful not to overread the financing as proof of technical superiority or market dominance. A large round indicates investor conviction, not independently verified performance. If Twelve Labs or its backers later publish benchmark claims around video search accuracy, retrieval quality, or enterprise adoption, those should be treated as vendor-reported unless independently validated.

The strongest evidence in this story is the funding event itself and Amazon’s reported participation. The weakest areas, at least from the materials available here, are product specifics and commercial traction. Those missing details matter because video AI can be expensive to train and serve, and enterprise demand depends heavily on integration quality and measurable accuracy.

What this means for builders and enterprise buyers

For AI builders, the funding highlights a practical opportunity: video is becoming a first-class input for applications, not just an afterthought attached to image or speech models. Teams building on Twelve Labs or competing platforms will likely focus on retrieval APIs, automatic tagging, clip extraction, summarization, moderation, and agent-like workflows that can act on video libraries.

For enterprise buyers, the main question is whether specialized video tooling delivers better economics and reliability than adding multimodal features from a general model provider. In some cases, a focused vendor may offer stronger indexing, lower operational friction, or domain-specific tuning for media-heavy workloads. In others, a broader provider may be “good enough,” especially if procurement prefers to consolidate on existing cloud or AI platforms.

This is where AI agents and workplace automation may eventually intersect with video infrastructure. Search is the first step; action is the next one. Once a system can reliably locate moments in video, businesses can begin automating downstream tasks such as assembling clips, routing incidents, checking policy compliance, or enriching a knowledge base. But those use cases depend on precision. A weak retrieval layer makes the rest of the stack brittle.

The raise also reinforces how enterprise AI is broadening beyond chat interfaces. Many organizations now want systems that can work across text, audio, images, and video inside business processes. In that sense, Twelve Labs is competing not only with other video startups, but with the direction of the broader multimodal market.

What to watch next

The next signals to monitor are straightforward. First, watch for an official Twelve Labs announcement that identifies the investors, the intended use of funds, and any roadmap priorities. Second, look for evidence of deeper ties, if any, between Twelve Labs and Amazon Web Services, especially around distribution, infrastructure, or joint enterprise go-to-market.

Third, product proof will matter more than funding headlines. Builders and buyers should watch for customer case studies, independent evaluations, API updates, pricing clarity, and latency or accuracy data that shows the platform can handle real production workloads. In multimodal AI, demos are easy to admire; dependable retrieval at scale is harder.

Finally, keep an eye on the competitive response from larger model vendors. If multimodal APIs from cloud platforms improve quickly enough, specialized players will need to show why their performance, tools, or economics justify a dedicated purchase.

Creati.ai perspective

This financing is best read as a bet on missing infrastructure, not just on one startup’s branding. Video remains a large and under-structured data source inside enterprises, and the company that helps turn it into searchable operational data could become deeply embedded in workflows. That is a stronger strategic position than many consumer-facing AI demos, but it also comes with tougher technical and economic demands.

For the market, the key takeaway is that multimodal AI is moving from novelty toward retrieval and operations. Twelve Labs now has the capital to try to own that layer for video. Whether it becomes a durable independent platform will depend less on fundraising momentum than on measurable product performance, integration depth, and whether specialized video AI can stay ahead of general-purpose multimodal systems.