How eToro and Google DeepMind Shipped One of the First End-to-End Veo 2 Brand Campaigns

Why eToro built a Veo 2 brand campaign on one model

In 2025 eToro partnered with Google to build what it described as one of the first end-to-end Veo 2 brand campaign, executed entirely on the generative video model from Google DeepMind. The work shipped as two 30-second films made for the Italian market, directed alongside AI-video specialist László Gaál with agency Marketing Arena. According to eToro's VP of Brand Marketing Yannay Politi, Veo 2 'proved to be the perfect fit for eToro's brand strategy, offering a next-level solution to produce compelling localized videos, providing realistic motion and high-quality output at a remarkable speed.'

AI video model selection is the discipline of matching a generative model to the job before you prompt, and eToro's pick of Veo 2 fit a brief that needed realistic motion and fast localized versions. Google's own documentation positions Veo 2 for text- and image-prompted generation with Content Credentials (C2PA) support, which matters for any brand putting synthetic footage into paid media because provenance can travel with the file from generation through delivery.

What makes this a useful case study is not the novelty of generative video itself but the completeness of the stack. eToro did not ship a single experimental clip; it put a paid, multi-channel campaign on air across YouTube, television, connected TV, and outdoor in Italy. That forces every downstream question, model fit, localization, direction, cost, and rights, into the open, which is exactly why it is worth dissecting for teams planning their own generative launches.

Localizing one concept for the Italian market

The campaign was built first for Italy, using local research to steer the creative. A survey by ACRI cited in the campaign coverage found that 28% of Italians under 44 save primarily for travel and leisure, 20% for durable goods, and only 10% for long-term wealth, so the films leaned into aspirational moments like driving through Tuscany or a wedding on Lake Como rather than abstract product messaging.

The AI video localization playbook locks a single master and splits the work into owned layers so each market version stays on brand, which is exactly how eToro's Italy cut was built from one concept. Localization here was not a separate reshoot but a variant of the same generated asset, and that is where model-led production earns its keep against a traditional multi-market shoot that would have rebuilt each film from scratch.

The Italy-first decision also reflects how generative campaigns travel. Once the master exists as a prompt-and-asset package, the marginal cost of a new locale is a set of reference images and a translated brief, not a new production call sheet. eToro signaled plans to expand into additional markets where it operates, and the asset structure is what makes that expansion credible rather than hypothetical.

The human direction behind the generated frames

For all the talk of autonomous generation, the campaign still ran through people. Gaál led the creative alongside eToro's marketing team and Marketing Arena, shaping shot intent, performance, and pacing rather than letting the model improvise every frame. That human layer is what kept the films from reading as generic AI footage and what made them defensible as a brand statement.

The human-core, AI-scaled creative model keeps a human art director in charge of concept while AI scales execution, and eToro's films relied on Gaál's direction plus agency scripting rather than fully autonomous generation. The takeaway for brand teams is that the model produces frames, but the director still owns the story, the performance, and the call on whether a take is good enough to ship.

This is also where the 2026 capability gap shows up. The 2026 CMO Barometer found that while 68% of CMOs name AI the defining topic of 2026, only 12% expect their agencies to lead on AI-specific skills—brands are keeping AI as their own strategic challenge rather than handing it to creative shops. eToro's answer was not to remove the human but to put the human earlier, at the concept and direction stage, where judgment is hardest to automate.

What the production economics look like

Model-led campaigns sit inside a fast-growing budget pool. The IAB projects U.S. digital video ad spending to surpass $80 billion in 2026—about $81.9 billion, up 11% year over year and nearly 20% faster than the total ad market—with digital taking more than 60% of all television and video dollars for the first time. Two-thirds of video buyers are live, testing, or planning agentic AI for campaigns in 2026, so the spend is moving toward AI-shaped production whether or not individual brands lead the way.

AI video production cost in 2026 puts real 2026 numbers on cost per usable clip, which is the metric that decides whether a model-led campaign like this beats a traditional shoot. When a 30-second cut can be iterated in hours instead of re-shot, the economics shift from hourly production cost to cost per usable clip, and that is the number procurement should be watching when it compares AI pipelines to legacy production.

The budget context also explains why a brand like eToro moved first. In a market where digital video spend is compounding and buyers are racing to adopt agentic tooling, the cost of being late to a working generative workflow is a loss of production leverage, not just a missed experiment. The campaign reads as a capabilities demonstration aimed at the market as much as at consumers.

Keeping a synthetic campaign rights-safe

A fully generated campaign still has to clear the same rights and disclosure bar as any other paid creative. Veo 2's support for Content Credentials (C2PA) means provenance can travel with the file, but the brand still owns the decision to label and disclose. eToro's campaign was framed as a statement about the future of marketing, which makes transparency part of the brand message rather than a fine-print afterthought bolted on at launch.

AI video commercial rights is the pre-flight checklist that keeps a generated cut compliant before it ships, covering platform license tiers, disclosure rules, and provenance. Treat it as a gate, not a retroactive fix: the cheapest place to catch a rights problem is before the media plan goes live, not after a platform pulls the creative for a missing label.

The rights conversation is sharpening across jurisdictions at the same time. Several markets now require synthetic-content disclosure on advertising, and platforms are building provenance checks into ingestion. A campaign that bakes disclosure and C2PA into the asset from the first generation is far easier to clear than one that treats compliance as a post-production step bolted on before the buy.

Five lessons for commercial video teams

The eToro and Veo 2 campaign is a clean reference point because it shows the whole stack at once: a model choice, a localization strategy, a human director, a budget context, and a rights gate. None of those pieces is optional, and the model is only one of them. Teams that copy only the generation step and skip the rest will ship footage that looks impressive in a cutdown and falls apart in review.

Start with the model fit, localize from a single master, keep a director in the loop, measure cost per usable clip, and clear rights before launch. The technology is now good enough to carry a real brand campaign, but the discipline is what makes it safe to put one on air, and that is the part no vendor can generate for you.

The throughline is that generative video changes where the work happens, not whether judgment is required. eToro's campaign moved the labor from the shoot to the brief, the direction, and the review, and it kept a person accountable at each of those gates. That is the operating model other brand teams should copy long before they copy the tooling.