Veo 3 was the standout announcement of Google I/O 2025: a video generation model that outputs not just footage but matching audio along with it, giving AI video a complete form factor for the first time. At launch it was reserved for the top-tier Google AI Ultra subscription, leaving users on the $19.99-per-month AI Pro plan with nothing but official demos. That changed on May 30.
As 9to5Google reported that day, Google opened Veo 3 to AI Pro subscribers in the Gemini app as a trial, on both Android and iOS, and expanded access in the Flow creation tool at the same time. The demand numbers were extraordinary. DeepMind CEO Demis Hassabis said users had generated millions of AI videos within days of launch, and the team spent the entire week just keeping the systems from collapsing under load.
From I/O exclusivity to Pro-tier trials
At I/O in late May 2025, Veo 3 was available only to Google AI Ultra subscribers, which left most paying Gemini users watching from the sidelines. The May 30 expansion changed that: AI Pro subscribers, at $19.99 per month, could try Veo 3 video generation in the Gemini app, on both mobile platforms, with Flow benefiting in parallel. Moving from exclusivity to a trial for the cheaper tier shows Google balancing its subscription ladder: let potential Ultra upgraders taste the model first, while the video capability of the Pro plan stops being theoretical. The trial also hints that Google is watching capacity, letting a subset of subscribers’ usage in to see how the infrastructure holds before deciding how much wider to open next. Given millions of videos generated in days, that staged approach is a reasonable choice.
Explosive demand: millions of videos in days
The Decoder reported on May 31, citing Hassabis, that Veo 3 users had produced millions of videos within just a few days of launch. Josh Woodward, VP at Google Labs and the Gemini app, described interest as “way, way, way more demand than we expected,” with the team spending the whole week keeping systems stable. Hassabis credited the infrastructure, chip, and SRE teams with keeping the TPUs from “melting”. Similarweb even measured a noticeable Veo 3 effect on DeepMind’s website traffic as curiosity drove visits to the model’s home. For a video generation model, that is a phenomenon-level start, and it explains why Google rushed to widen access again and again. The frenzy also doubled as marketing: viral clips showcasing synchronized audio gave the model a reach no campaign could buy, and handed competitors a clear demand signal for the entire AI video category.
Geographic reach: over 70 countries, Europe still waiting
According to The Decoder, surging demand pushed Google to expand Veo 3 to 71 countries, with the UK added recently, while the EU remained unavailable pending a rollout Google said was on the way. 9to5Google separately noted that the AI Ultra subscription itself had reached 73 countries. The two figures differ slightly, one counting markets where the model is available and the other the subscription’s footprint, but the direction is the same: Veo 3’s reach grew fast, while European users, held back by rollout pacing, had to wait longer and experienced a visible gap between markets.
Watermarks and content labeling
Alongside the wider rollout, Google attached content labeling to Veo 3: every generated video now carries a visible watermark, with one exception, as AI Ultra users working in Flow can get watermark-free results. Google described this as an initial step toward clearly marking AI-generated content and said it was working to broaden access to its SynthID Detector for identifying synthetic media. The Decoder questioned how much practical force the scheme really has when users can bypass the watermark simply by upgrading to Ultra. How watermarking policy coexists with commercial incentives remained the open question of the expansion. For creators, the watermark also shapes judgments about remixing and commercial use, and the threshold for getting watermark-free output effectively becomes one of Ultra’s differentiators, binding subscription tiers more tightly to content rights.
What it meant for creators and developers
Two things made the Veo 3 surge worth remembering. First, generating picture and audio in one pass gave AI video the complete experience social platforms demand, changing the cost structure of short-form content creation, and users immediately showcased the capability across social media. Second, the demand curve proved video generation was real usage rather than a technical demo, leaving Google to keep calibrating compute allocation, watermark policy, and subscription tiers. Looking back, the late-May 2025 expansion marks the moment AI video crossed from demo to production-scale application, and it foreshadowed the competition over synchronized audio-and-video generation that followed. Pricing tiers, compute budgets, and labeling rules would keep moving together from here.
Sources
- 9to5Google: Gemini app Pro users getting Veo 3 trial, AI Ultra now in 73 countries
- The Decoder: Google says Veo 3 users have generated millions of AI videos in just a few days
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.