1.24 Billion AI Videos in 30 Days: What $0.05/Second Means for Your Ad Creative

xAI dropped a number I haven’t been able to put down.

1.24 billion videos generated in 30 days.

Not a product demo. Not a capped beta. The first full month of Grok Imagine, running wide open.

For scale: YouTube takes in about 500 hours of video every minute. Grok showed up in month one and matched that, at a fraction of the cost.

If you run paid ads, that number should change how you plan your next quarter of creative. This piece sits inside how I think about AI content marketing: where the creative work is heading, and who gets there before the rest of the market wakes up.

(Video: $0.75 Ad Videos? Grok AI Just Did It, MarketingAlec on YouTube)

The number that actually counts: $0.05

At the time of the original post, Grok’s Aurora API ran about $0.05 per second of generated video.

A 15-second ad clip came to roughly $0.75. Seventy-five cents, with native audio, multiple aspect ratios, and 720p output sized for social.

Stack that against what the other tools were quoting then:

Veo 3 (Google): around $0.20/sec, roughly 4x the price
Sora 2 Pro (OpenAI): around $0.25/sec, roughly 5x the price
Traditional production: $3,000 to $8,000 for a single 60-second spot

(AI video pricing moves fast, so read those as the snapshot from when this ran, not today’s rate card. Check the current API pages before you budget anything.)

xAI didn’t just enter the AI video market. They repriced it.

What cheap video actually buys you

The win isn’t “make one video for less money.”

It’s test velocity.

Most brands run 2 to 4 creative variants per campaign, because production is slow and pricey. At under a dollar a clip, you can test 20 variants for what you’d spend on lunch. That changes the whole shape of how you buy media.

Instead of: “Which of these 3 ads should we run?”

You run: “Generate 20 hooks, test them all this week, find the 2 winners, scale those.”

This is what I’m watching the sharper brands do right now. They ship video creative at the same pace they ship static ads.

The practical playbook

1. Script with AI

Open Claude or ChatGPT.

Prompt:

You are an advertiser specializing in short-form social video ads. Your objective is to write 5 different 10-second video ad scripts for [product/service] that each stop the scroll with a different hook type, by following direct response principles for social-first video.

Each script must:

Open with a unique hook (use one of: bold claim, question, pattern interrupt, pain point, social proof)

Stay under 30 words total (10 seconds of spoken audio)

End with one clear, specific CTA

Use casual, direct language, no corporate speak, no jargon

Format each script exactly like this:

Script [number]: [hook type]

VISUAL: [one-line scene description for AI video generation]

AUDIO: “[the spoken script]”

CTA: “[the closing call to action]”

Focus on actionable messaging. Each script should feel like a different ad, not a variation of the same one. Confirm that each hook type is distinct and each CTA is specific.

2. Generate with Grok Aurora

Feed each script to the API. At roughly $0.05/sec for 10-second clips, you just produced 5 complete video ads, audio included, for about $2.50.

3. Test all 5

Run them as a split test. Same budget, different creative. Let the data pick the winner.

4. Iterate

Take the winner. Generate 5 variations on that concept. Test again. In two weeks you’ll know more about what works for your audience than most brands learn from a full year of quarterly creative refreshes.

Aurora is good. It’s not magic. Veo is better.

It’s built for short social clips: 15 seconds max, 720p. You’re not producing a brand film or a broadcast spot here. Anything that needs real faces, your actual product, or live action still wants a human editor and a few different AI models working together.

And the best-performing ads on Meta and TikTok right now aren’t cinematic anyway. They’re fast, direct, and a little rough around the edges. Aurora’s output fits that look perfectly.

(Video: Turn Your Product Images Into AI Videos with ScaledOn, ScaledOn on YouTube)

Need more video content?

Our AI-assisted video workflow turns product photos and brand assets into ready-to-post videos, so you can make more of them, faster.

Turn your product photos into videos with ScaledOn: https://scaledon.com/ai-video

Why this moment matters

1.24 billion videos in month one tells you one thing plainly: AI video isn’t a niche toy anymore. It hit mass adoption.

Which means your competitors are sorting this out too.

The brands that build a test-first creative workflow now, before it turns into table stakes, get a real head start. The gap between “we test 3 creatives a quarter” and “we test 30 a week” is enormous in terms of what you learn about your audience. If you want the wider view of where this fits in your content engine, that’s the AI content marketing map, and it connects up to the full AI marketing picture.

Your next ad video might cost less than a dollar.

So make a lot of them.

Cheers,

Alec

Your ad budget just stopped being the bottleneck

When a 15-second clip costs less than a coffee, the brands that win aren’t the ones with the biggest production budget. They’re the ones treating every clip as a cheap experiment and reading the results faster than everyone else. Every Friday I send the AI creative workflows I’m running for real: the exact prompts, the tools I reach for, the tests that landed and the ones that flopped.

Subscribe free and grab the test-first playbook before your competitors build it.