When AI Giants Discover They Need a Human Face

I left Nokia in 2007 after the launch of the iPhone. Not because the iPhone launched, but because of their reaction to it at the time. "people hate touchscreens" and "only a 1.3 megapixel camera, ha!" were some of the internal, even celebratory, refrains at the time from an engineer-led company, and you could smell the doom.

I went to Orange, where despite being a mobile operator we never showed a device. We weren't selling bandwidth, we were selling the ability to send your mum a photo of her grandchild. We didn't lead with network coverage maps, we led with "The future's bright, the future's Orange", a promise about optimism and possibility, not radio frequency spectrum.

Same industry vertical, very different approaches. One engineer-led. The other led by engineers plus other humans. Marketers like me who make the truth interesting.

The technology industry has always suffered from the same delusion: that product feature superiority alone creates sustainable competitive advantage. It doesn't. It never has. And this past month, OpenAI and Anthropic are learning this lesson the expensive way.

Within one week of each other in late last month, both AI giants launched their first major consumer-facing brand campaigns. OpenAI spent millions on 30-second spots shot on 35mm film for NFL Primetime. Anthropic partnered with agency Mother for a 90-second manifesto about problem-solving that ran across Netflix, Hulu, and the New York Times.

For anyone else who's lived through technology cycles, from mobile phones to smartphones, from broadband to streaming, this moment is instantly recognisable. It's the moment when product-led growth hits a wall, when commoditisation arrives faster than anyone predicted, and when companies scrambling for differentiation rediscover the ancient art of brand building.

But here's what the trade press isn't discussing: these campaigns aren't really about consumers at all.

Let me give you the numbers that should terrify both companies' boards, and that AdWeek, The Drum, and TechCrunch are politely ignoring.

ChatGPT has 700 million weekly users, four times higher than last year, with 200 million added in just the last six months. That's staggering adoption. Industry analysts celebrate this growth as validation of AI's inevitable dominance. Trade publications frame it as a success story.

It's not. It's the largest conversion failure in technology history.

Only 3-5% of ChatGPT's weekly users pay for subscriptions. Read that again. OpenAI has achieved near-universal awareness, massive adoption, and utterly failed at monetisation. Meanwhile, growth in North America and Europe, their primary markets, has flatlined. To celebrate growth when you are giving away something for free that improves people's lives is weird.

AI companies are burning through billions in venture capital to subsidise free users they cannot convert to paying customers at scale. When that subsidy ends, and it must, the entire industry faces a reckoning.

The brand campaigns aren't about building preference. They're expensive signals to investors that management recognises the conversion crisis and is doing something that looks like a grown-up company attempting to solve it.

Anthropic's challenge is different but equally existential. The company has grown from fewer than 1,000 business customers two years ago to over 300,000 today. Its revenue run-rate jumped from $1 billion at the start of 2025 to over $5 billion by August which is explosive growth by any standard.

Yet despite this, it remains almost invisible to consumers, dominated entirely by OpenAI in public consciousness. More critically, those impressive customer numbers mask troubling unit economics: $5 billion revenue ÷ 300,000 customers = $16,667 annual revenue per customer. But the top 1% of customers likely generate 50%+ of revenue, meaning the other 297,000 customers average just $2,500-5,000 annually. These aren't sustainable enterprise contracts, they're prosumer subscriptions at scale, with consumer-level churn risk.

Both companies raised a combined $44.5 billion in Q1 2025 alone, nearly half of all US venture capital that quarter. They're sitting on $183 billion (Anthropic) and $300 billion (OpenAI) valuations. And both are discovering what every technology company eventually learns: when you can't differentiate on product, you've got to differentiate on brand.

Watch the campaigns and you'll see two companies making fundamentally different strategic errors in work that is otherwise beautiful.

OpenAI's "Everyday Magic" approach shows a man achieving his first pull-up after following ChatGPT's eight-week training plan. Another spot features someone cooking an impressive dinner for a date using a ChatGPT recipe. It's warm, human, aspirational and shot beautifully on 35mm film with indie music from Perfume Genius and Simple Minds.

The creative is excellent. The strategy is confused.

OpenAI generates roughly 80% of its revenue from ChatGPT Plus consumer subscriptions. They have massive awareness and usage, but conversion is their crisis. They need people to pay $20/month for the premium tier. Yet these ads do nothing to articulate why the free tier isn't enough. They celebrate usage without driving the behaviour that matters: upgrading.

This is a strategic failure that reveals how desperately OpenAI wants to look like they're addressing their business model problem without actually addressing it.

Anthropic's "Keep Thinking" campaign positions Claude as "the AI for problem solvers—those who see AI not as a shortcut, but as a thinking partner." The 90-second film features montages of people coding, fixing bikes, making art, doing marine conservation. The message: there's never been a better time to have a problem.

Again, the creative is sophisticated. The strategy misses the mark.

Anthropic derives 80% of its revenue from API usage by developers and enterprise customers. They don't have a consumer awareness problem; they have an enterprise credibility gap and a unit economics challenge. Yet this campaign chases consumer emotional connection rather than demonstrating the technical reliability, security posture, and integration capabilities that CIOs actually care about, or solving the fundamental problem that most of their customers don't spend enough to justify the acquisition and support costs.

Both campaigns make the same fundamental mistake: they're trying to look like mature brands before they've solved their actual business model challenges.

I posit then that these campaigns aren't primarily about acquiring customers. They're about signaling maturity to the investors who've poured tens of billions into companies with unproven (and possibly unprovable) business models.

After leading marketing at Publicis and watching the due diligence process for major transformations, I recognise this pattern. Professional marketing infrastructure (hired CMOs, retained agencies, measured media strategies) these are boxes that must be ticked before public market scrutiny.

OpenAI appointed Kate Rouch, their first CMO, in December 2024 from Coinbase. Elke Karskens joined six months ago to build their international marketing function. Anthropic brought in Andrew Stirk as Head of Brand Marketing. Both companies appointed media agencies of record this year (PHD for OpenAI, Initiative for Anthropic.)

This is pre-IPO theatre.

The timing tells you everything. These launches came as both companies closed massive funding rounds: $13 billion for Anthropic, $8.3 billion for OpenAI in recent months. When you raise that kind of capital while sitting on conversion rates that would be considered catastrophic in any other industry, investors expect you to spend some of it looking like a company worthy of its valuation.

The unspoken question these campaigns try to answer: "Can we build brand equity and pricing power before the venture subsidy runs out and we have to charge what our service actually costs?"

The answer matters because the math is brutal. OpenAI likely spends $5-7 billion annually on compute alone, before R&D, salaries, and infrastructure. With 665 million free users, they're subsidiisng roughly $2.87 billion in annual compute costs for users who will never pay them a penny.

I've watched this movie before. Technology companies that invest heavily in brand building when their business models are uncertain typically follow one of two paths:

Path 1: They use the campaigns as cover for fundamental business model pivots. The advertising buys them time and creates the perception of confidence while they fix the underlying economics. This sometimes works if the pivot happens fast enough.

Path 2: They double down on brand awareness as a substitute for solving hard strategic problems. The campaigns get bigger, the creative gets more awarded, and the business model gets worse. This never works.

For marketing leaders, the lesson is clear: brand building is essential, but it's not a substitute for business model clarity. The strongest brands are built on clear value propositions, not vague inspiration. And when your conversion problem is existential, no amount of beautiful advertising will save you.

Brandflow is written by Justin Billingsley, who has spent his career on all three sides of the industry's table: senior client, global agency leader, technology founder. First published 8 October 2025 in the Brandflow newsletter on LinkedIn.