Japan's ai& launches ai& video with 12-second generation

ai& launched ai& video on October 22, 2026, turning Japanese text, photos and existing clips into new footage.

The company styles its name ai&, a lowercase "ai" closed by an ampersand that stands in for "and", so the name reads "ai and". Its corporate name is ai& Inc. and its site is aiand.com. ai& announced the service in a post on its own newsroom, dated October 22, 2026, and in a press release the same day.

ai& video takes three kinds of input. A user describes who is in a scene, where it happens and what happens, in plain Japanese. They can animate a single photograph, using its subject and lighting as a starting point, or continue an existing clip. The service runs in a browser and through an API, so a production tool or an app can call it.

ai& says generation can take as little as 12 seconds on its own infrastructure in Japan. The company marks the figure with an asterisk: it is a measurement under specified conditions, and actual times vary with the model, the settings and the length of the video. On supported models, ai& video also generates audio alongside the video.

Pricing is pay-as-you-go and varies by model, with no monthly subscription. In a technical blog posted the same day, ai& lists 768p launch pricing per generated second: H3 Fast at $0.0125 and H3 Quality at $0.055 for text or first-and-last-frame input, with reference input at $0.075. The list notes that USD and JPY are separate list prices. Enterprise customers can order additional training on characters and footage they hold rights to, which produces a private model in an environment other users cannot reach. ai& says material uploaded for standard generation is never used to train models. An in-house moderation system, built to read Japanese prompts, screens output for inappropriate or potentially infringing use.

ai& is a Yokohama-based AI infrastructure company founded by David Bennett, formerly chief customer officer at Tenstorrent and former chief executive and president of NEC Personal Computers / Lenovo Japan, and Shimpei Hara, also formerly of Tenstorrent. It launched on March 25, 2026 with $50 million in seed funding and more than $2 billion in committed data center capital, and its ai& inference platform, which serves models for language and scientific research, launched on June 29, 2026. ai& video extends that platform to video generation.

Shimpei Hara, the co-founder and president, said in the release: "We want to build an environment where waiting time and cost never make you stop."

A technical blog ai& posted the same day describes running the H3 video model on NVIDIA, AMD and Tenstorrent hardware and the tradeoffs between speed, cost and quality. Pay-as-you-go pricing without a subscription could suit creators who make video occasionally, if the per-generation cost lands where ai& sets it.

ai& video is live now at aiand.com/video, in a browser or through the API.