Clipto disclosed a $250M-plus valuation in January; TechCrunch reports a $15M round

Why it matters

Kang is betting that personal AI memory can become a standalone software layer. Clipto's claimed profitability gives that thesis weight, while platform competition puts its durability to the test.

Hongwen "Henry" Kang built Clipto around a contrarian AI bet: the growing pile of files people already own may be a better business than generating another pile. In its January 5 funding release, Clipto claimed it generates more than $15 million in annual recurring revenue and disclosed financing at a valuation above $250 million. The company did not state the latest round's size, while TechCrunch's August 31 profile described it as a $15 million round. ION Analytics reported that Kang said the company reached profitability after three months.

The financing activity began in 2025, before Clipto disclosed the valuation in January. Clipto said it had completed Pre-A, Pre-A+ and Pre-A++ rounds beginning in July 2025. ION Analytics reported that Clipto raised approximately $20 million from investors during 2025. The January release specifically named 522 Ventures. ION Analytics associated Palm Drive Capital, EnvisionX Capital, HSG, GL Ventures and 522 Ventures, as well as individual investors Hans Tung and Lu Zhang, with the 2025 financing.

Kang's pitch centers on indexing media people and businesses already possess. Clipto indexes video, audio, images, meetings and documents, then lets users search for a spoken phrase, person, action or scene and jump to the relevant moment. Clipto describes its Mac and Windows products as local-first, while its current product page says its iPhone, Android and web versions use cloud processing.

A decade-long loop back to understanding

Kang has spent much of his career working on variations of the same technical problem. At Carnegie Mellon University's Robotics Institute, he studied how machines could identify and remember objects in everyday environments. He completed a robotics Ph.D. in December 2012 under computer-vision researchers Takeo Kanade and Martial Hebert; his dissertation was titled "Object Instance Discovery from Scenes of Daily Living."

He then applied computer vision to consumer products and video creation. Kang founded an AI fashion-discovery business before building ZenVideo, a video-creation platform that Tencent acquired in 2020, he told TechCrunch. Kang stayed at Tencent working on content-generation products, then left with members of his earlier founding team in November 2023 to start Clipto.

That history explains Clipto's positioning. Kang spent years helping machines generate media, then returned to the problem of understanding and retrieving it. In a June interview with 36Kr, he described cloud models as a global brain and edge AI as a personal memory. Clipto is trying to make that memory durable enough that users keep it even as the AI agents accessing it change.

The local-first wedge

Clipto's January release describes use cases in healthcare, legal work, marketing, education and research, extending the product beyond media professionals. That breadth matters because creator software alone would leave Clipto exposed to editing suites that already own the workflow.

Clipto is recasting its index as a memory layer that other AI products can query. Clipto recently added Model Context Protocol support, allowing tools such as ChatGPT, Claude and Cursor to retrieve information from an indexed library after the user authorizes access. Its current product page also lists integrations with Premiere Pro and DaVinci Resolve, alongside Mac software and a Windows beta.

Local processing gives Kang a defensible technical problem to solve. Large media collections are expensive and slow to upload, while legal, medical and corporate recordings can carry obvious privacy constraints. Clipto says it has built more than 10 proprietary multimodal models plus an edge-inference and compute-scheduling system. Kang told 36Kr that an M5 MacBook Pro could index about 2 terabytes of video in 24 hours, work he estimated would cost roughly $400 in the cloud. Those performance and cost figures are Clipto's benchmarks, rather than an independent test.

The revenue claims need their denominator

In its January funding announcement, Clipto claimed more than 10 million users across more than 100 countries. Clipto also says it has hundreds of thousands of paying subscribers. Kang gave TechCrunch neither a precise subscriber count nor average revenue per customer. He said Clipto reached $15 million in ARR at the beginning of 2026 and remains profitable on a net-income basis. All of those operating figures come from Kang or Clipto.

The subscriber and revenue figures do not line up neatly with Clipto's current US list prices. The pricing page lists Clipto Pro at $24.99 monthly, with an annual plan equivalent to $12.49 a month. Even 200,000 subscribers paying the annual-plan rate would represent nearly $30 million in annualized revenue before discounts. Regional pricing, older plans, promotions or a broad definition of paying subscribers could account for the difference, but Clipto's disclosed figures do not establish which explanation applies.

The fundraising history offers limited detail. An ION Analytics report said Clipto raised approximately $20 million during 2025 and was valued above $250 million. Clipto's January announcement grouped several rounds together, supplied no round-by-round amounts and did not disclose the size of its latest financing; TechCrunch later reported that latest round as $15 million. Investors repeatedly financed Kang's local-AI thesis during 2025, and Clipto said the capital would fund model development, consumer-device infrastructure and additional agent integrations.

A standalone product in platform territory

Clipto is building in territory that large platforms already treat as a feature. Adobe's Premiere can analyze footage with on-device models and search imagery, dialogue and metadata using natural language.

Specialists are also well funded. TwelveLabs, an enterprise provider of video-understanding models and APIs, announced a $100 million Series B on July 1. Clipto is pursuing a different customer and distribution model: a consumer-style desktop application that indexes files across formats, uses on-device processing for its primary memory layer and exposes selected material to outside agents.

Kang's challenge is to turn that cross-application memory into something users regard as infrastructure rather than a replaceable search utility. The reported revenue and profitability suggest customers are already paying for the distinction, though Clipto's self-reported metrics leave the scale of that demand less settled than its valuation above $250 million implies. The funding gives Kang room to prove that personal media memory can remain an independent product even as Adobe keeps absorbing its most visible features.