An analysis of more than 1.5 million videos made on Pictory in 2026 finds AI-native video features cluster by geography — Oregon creators generate AI images at roughly 9x the US average, and Pennsylvania leads the country in AI avatar adoption.
2026-08-17 · by Moe Ameen
Pictory, an AI video-creation platform, analyzed aggregated, anonymized data from more than 1.5 million videos made on its platform in 2026 and released its "2026 State of the AI Video-Creation Industry Report" in June 2026. To compare places fairly regardless of population, Pictory normalized the metrics per 1,000 users — so the figures describe how heavily the average creator in a state or country uses a given feature, not total output.
The clearest state-level findings are in the AI-native features. Oregon leads the US in AI image generation, with creators generating images at 1,215 per 1,000 users — roughly 9x the national average. Pennsylvania leads the US in AI avatar adoption, at 638 avatar uses per 1,000 users, about 4x California's rate and 6x New York's. Voiceover adoption concentrates internationally rather than in any one US state: Denmark leads the world, using custom voiceovers at nearly 7x the US rate. Pictory also reports that professional users run URL-to-video workflows at 9x the rate of personal users, education and training teams use PowerPoint-to-video 4x more often than YouTube creators, and video length varies by creator type — sales and marketing teams average about 1.7 minutes per video versus 3.9 minutes for YouTube creators. Mac users adopt avatars, AI images, and voiceovers at consistently higher rates than other users.
The timing pattern is consistent across the dataset. The busiest hour for AI video creation globally is 9pm local time, and roughly 35% of all daily video creation happens between 7pm and 1am — well outside conventional working hours, pointing to evening, batch behavior rather than in-the-workday production.
Two caveats keep this honest. The data comes from a single platform and reflects Pictory's own user base and product mix (marketing, education, and small-business use), so it is directional rather than a census of all AI video made in America. And per-1,000-user rates can look dramatic precisely because they measure concentration among that state's Pictory users, not total volume. The durable, cross-cut finding is the one Pictory highlights: AI-native formats — generated images, synthetic voiceovers, and avatars — have moved from novelty to everyday professional workflow, and where a creator sits predicts which of those they reach for.
Read the study as a shopping list of what creators are actually making, and the mismatch with a single-purpose tool becomes obvious. The features driving the numbers — AI avatars (Pennsylvania leads at 638 per 1,000 users), generated images (Oregon leads at ~9x the US average), and synthetic voiceovers — are separate products or add-ons in most stacks. [Kompozy](/) generates all of them from one engine: avatar and talking-head video via [Persona Shorts and Persona HeyGen](/glossary/output-buckets) with native voiceover, face-locked [AI images and Persona Photos](/glossary/output-buckets), plus carousels, blogs, and newsletters that a text-to-video montage tool never touches. So instead of copying the "leading region" workflow by bolting an avatar app onto an image app onto a voiceover app, a creator in any state runs the whole AI-native format set — the ones the data says are winning — inside a single [Persona Brief](/glossary/persona-brief) that keeps voice and face consistent across every output.
The timing finding is where Kompozy quietly does the heavy lifting. The report shows creators building at 9pm, with ~35% of creation happening between 7pm and 1am — batch behavior, done off-hours, on the maker's schedule. That's only a problem if creation time and publish time are the same clock. Point Kompozy at a batch built late in the evening and its [scheduler and Autopilot](/glossary/autopilot) fan each piece out to the eight primary social platforms plus blog and email at the right time for each audience, behind a per-post review gate — so the evening build habit the study documents becomes a full week of well-timed posts without you being online to push them. And because the geography data suggests adoption is a behavior anyone can copy rather than a fixed regional advantage, the practical move is to stop [repurposing](/glossary/content-repurposing) one clip at a time and start generating the full format mix on a cadence. If you want the news-cycle version too, feed the study straight into Kompozy and it becomes a blog explainer of the state map, a carousel of the leading states per feature, and quote cards of the standout figures — published everywhere in a day while the report is still being passed around.
Oregon leads the US in AI image generation, with creators generating images at 1,215 per 1,000 users — roughly 9x the national average. Pennsylvania leads the US in AI avatar adoption, at 638 avatar uses per 1,000 users, about 4x California's rate. Because Pictory normalized the data per 1,000 users rather than by total output, these figures describe how heavily the average creator in a state uses each feature, not which state produced the most videos overall.
Oregon dominates AI image generation at 1,215 images per 1,000 users (~9x the US average), and Pennsylvania leads AI avatar adoption at 638 per 1,000 users (about 4x California's rate and 6x New York's). Voiceover adoption concentrated internationally rather than in a single US state — Pictory found Denmark leading the world at nearly 7x the US rate.
Across the global dataset, 9pm local time is the busiest hour and roughly 35% of all daily video creation happens between 7pm and 1am — meaning most creation is batched into the evenings, outside standard working hours.
Treat them as directional. The data comes from a single platform (Pictory) and reflects its own user base and product mix, so it is not a census of all AI video made in the US, and per-1,000-user rates measure concentration among that state's users rather than total volume. The robust takeaway is the trend — AI-native formats like avatars, images, and voiceovers have become everyday professional workflows, and adoption clusters by creator behavior.