Google’s new update to its AI video model adds vertical video into the mix, and that one change could have an oversized effect on TikTok, Reels, and Shorts. Added support for 9:16 output in Veo 3 via the Gemini API means developers can create mobile-native highlights without ugly cropping or post-production bodge jobs.
It’s a simple change with high-level consequences: Creators and product teams can now create AI videos that match social feeds as users experience them, to the pixel. For a format that thrives and dies on friction, that adds up to a removal of the most common formatting friction.
Why 9:16 matters for AI video
Vertical video is not a preference; it’s the default for so much social consumption. TikTok now has more than a billion monthly users worldwide, YouTube claims that Shorts is watched by over two billion logged-in users each month, and Meta has said in earnings calls that more than 200 billion Reels are played daily. That is the sea that Veo 3 is wading in to.
Until now, a lot of the AI-generated clips were born in landscape and then cropped, which can affect framing, legibility of on-screen text and perceived quality. Native 9:16 content maintains all proposed composition, (foreground subjects take full scale), and rather than a down-rezzed, stitched together, or reflowed awkwardly, the proposed motion paths were tailored to the vertical viewport – which can often times mean better retention and less creative retakes.
What developers get today
Developers can request Veo 3 outputs via the Gemini API in a 9:16 aspect ratio, matching TikTok, Instagram Reels, and YouTube Shorts. That means less letterboxing or multi-pass reframing, and your assets are “social-ready” straight out of the render.
Synchronized Audio along with Video also can be processed using Veo 3. Rather than sewing in stock music or voice-over after-the-fact, teams can prototype narrative beats and pacing in one prompt-driven pass, then rework shots or sound design as necessary. Fewer steps for anyone building automation around social publishing means faster iteration and a lower cost per concept test.
Will consumer tools follow?
Developers who use the API get it first in a vertical format. Google is already surfacing Veo features in consumer-facing products — such as converting photos into brief clips inside the Gemini app and Google Photos — though it hasn’t said when a straightforward “vertical” toggle might arrive for everyday folks. If and when that shift comes soon, look for an explosion of AI-native vertical content hatched outside pro-production studios.
In the meantime, anticipate third-party tools and startups to package the API and provide one-click templates for common social situations: talking-head explainers, street vox pops, product demos and travel reels.
The sooner such wrappers come, the faster this feature will reach creator workflows at scale.
What this means for feeds and brands
This reduced the bar of scale due to its native vertical generation. The sort of AI-driven “man-on-the-street” interviews and hyperreal creator clips that already circulate can more easily be produced from the outset in the proper format, thereby inviting more experimentation and more uploads. For brands, it opens up quicker A/B testing on TikTok’s Ads Manager and Reels placements: many more variations, same budget.
There is a flip side. A vertical AI content frictionless pipeline could quicken the process of feed saturation. Platforms will rely on ranking signals — watch time, replays, user feedback — to weed out low-quality outputs. For creators, that will mean differentiation will depend on concept strength and editing discipline, not just access to a model.
Safety, labeling, and authenticity
As it becomes easier to generate, provenance becomes key. Google DeepMind has pushed SynthID, a watermark for AI media, and leading platforms are experimenting with content credentials through the Coalition for Content Provenance and Authenticity. TikTok and Meta also mandated labeling for AI-generated content that has been altered to look real. Enforcement is likely to tighten as vertical AI videos grow in popularity and as regulators turn up the heat on disclosure standards.
For organizations deploying Veo 3, adding automated labeling, caption disclosure and human review would help reduce reputational risk for noisy or sensitive situations where synthetic footage could easily mislead if not labeled as such.
The competitive backdrop
Competitors such as Runway, Pika and Luma have backed vertical outputs, and OpenAI’s Sora demos have shown a cinematic range although access is restricted. The Google move brings Veo 3 in line at the format level and then it’s just taking advantage of positioning: the Gemini API hooks into the larger Google developer ecosystem, to Android surfaces and perhaps advertising workflows.
Bottom line: vertical support is what will turn Veo 3 from an impressive model into a more usable one for the platforms where attention actually resides. For the time being it’s a victory for developers, who can rub their hands and churn out AI-native clips for the thumb-scroll.
