There’s a question nobody in Silicon Valley really wants to answer right now. It goes something like this: if you remove the youngest, most engaged, most algorithmically valuable users from your platform, what happens to the war chest funding your AI future?
It’s a question that’s becoming increasingly urgent. Australia moved first, rolling out a nationwide social media ban for under-16s in December 2025. Within the first month, platforms had already removed 4.7 million accounts, with Meta alone shutting over 500,000. The UK isn’t far behind. In January 2026, the House of Lords backed an amendment to the Children’s Wellbeing and Schools Bill that would ban under-16s from social media, and the government has launched a formal consultation with results expected by summer. France, Spain, Germany and Norway are all moving in the same direction. This isn’t a fringe policy conversation anymore. It’s a global legislative wave, and big tech is standing right in its path.
So let’s talk about the money.
The Numbers Are Larger Than You Think
In 2023, Harvard researchers published the first serious attempt to quantify what youth users are actually worth to social media platforms. Their findings were striking. Across just six platforms (Facebook, Instagram, Snapchat, TikTok, X and YouTube) the annual advertising revenue generated from US-based users under 18 came to nearly $11 billion. That’s in the US alone.
Break it down by platform and it gets even more revealing. Snapchat derived the greatest share of its overall ad revenue from users under 18 at 41%, followed by TikTok at 35%, YouTube at 27%, and Instagram at 16%. Instagram generated around $4 billion from teens aged 13-17, TikTok around $2 billion, and YouTube $1.2 billion, again, just from US adolescents.
Think about that for a moment. Snapchat, a platform whose entire identity is built around the fleeting, irreverent energy of youth, generates over four in every ten advertising dollars from people who aren’t yet legally adults. For TikTok, it’s more than a third of the business. These aren’t marginal users. They are the product.
And the real figure is almost certainly higher. These estimates are from 2022. Platform revenues have grown significantly since, and the researchers themselves acknowledged they were working from projections because, critically, platforms don’t disclose revenue by age group. They have every reason not to.
Why Young Users Are Worth So Much
It isn’t just about volume, though the numbers are significant. Teens are unusually valuable to social media platforms for several structural reasons.
First, they are disproportionately heavy users. Pew Research found in late 2024 that nearly half of US teens report being online “almost constantly”, with the majority visiting TikTok and YouTube daily. Heavy usage means more impressions, which means more ad revenue, but it also means more behavioural data.
This is the second and more consequential point. Young users generate enormous amounts of high-quality training signal. Every scroll, pause, share, rewatch and comment is a data point. These interactions feed the recommendation algorithms that make platforms sticky, and those same algorithms, refined on billions of micro-behaviours, form the technical bedrock on which AI content systems, personalisation engines and generative tools are being built. The teenager doom-scrolling TikTok at 11pm isn’t just an ad revenue opportunity. They are, functionally, an unpaid participant in an enormous machine learning experiment.
Remove that cohort from the equation and you lose both revenue streams simultaneously.
The Snapchat Problem and What It Reveals
Snapchat’s position is perhaps the most instructive case study here. Around 20% of its audience is under 18, a significantly higher share than Facebook, Instagram or TikTok, which may explain why the company has pushed back against the Australian ban and chose to settle a child social media addiction lawsuit while Meta, ByteDance and YouTube plan to go to trial.
Snap has been scrambling to diversify. Its subscription service, Snapchat+, recently crossed a $1 billion annualised revenue run rate, and the company is investing heavily in augmented reality. But these moves are partly defensive, a hedge against the day regulators remove the revenue base it has built on younger users. The pivot tells you everything about how seriously the company is taking the threat.
Meta and TikTok are in a structurally stronger position, with their teen revenue representing a smaller proportion of their overall business in relative terms. But “smaller proportion” is relative. Instagram’s $4 billion in teen ad revenue is still $4 billion. A meaningful chunk of that is likely in the 13-15 age bracket that a ban would directly target.
And What Does This Have to Do with AI?
Here’s where the conversation gets genuinely interesting, and where we think the industry has been strangely quiet.
The major social media players are not just social media companies anymore. Meta is spending tens of billions annually on AI infrastructure and research. Google (YouTube’s parent) is in an existential AI race with its Gemini models and is burning capital on compute at a rate that would have seemed implausible five years ago. TikTok’s parent ByteDance is one of the most sophisticated AI companies on the planet. Snap is betting its future on AI-powered AR.
All of that investment needs to be funded from somewhere. And it’s funded, primarily, by advertising revenue.
The connection between social media profitability and AI development pace is more direct than it might appear. When Meta reports strong ad revenue, it frees up capital for its AI research teams. When Snap’s revenue stagnates, as it has, it faces harder choices about where to cut. A significant reduction in the addressable advertising audience doesn’t just affect quarterly earnings. It compresses the innovation budget.
Consider: if a coordinated international ban on under-16s reduced major platform revenues by even 10-15% globally, we could be talking about tens of billions of pounds in annual revenue across the sector. That is real money that would otherwise flow into GPU clusters, AI research labs, and the talent wars that determine who leads in the next generation of AI development.
The Counter-Arguments Are Real, Too
We should be honest about the complexity here, because this isn’t a clean narrative.
The most obvious pushback is that the major AI labs, Anthropic, OpenAI, DeepMind, Mistral, are largely not dependent on social media advertising revenue. Their funding comes from venture capital, corporate investment and API revenue. A hit to Meta’s ad business doesn’t directly slow down GPT-5 or Claude.
There’s also the question of whether bans actually work. Australia’s experience is already revealing workarounds. VPN downloads surged almost immediately, and technically adept teenagers are migrating to smaller, less regulated platforms. If under-16s simply shift their attention elsewhere, the data and engagement doesn’t disappear; it just moves somewhere harder to measure and regulate.
And it’s worth noting the perverse incentive the ban might create. If platforms lose ad revenue from younger cohorts, they may actually accelerate their pivot toward subscription models and AI-driven monetisation of adult users, potentially speeding up certain kinds of AI product development rather than slowing them down.
The Bigger Question
But here’s the provocation we want to leave you with.
The social media ban debate is being framed almost entirely around child welfare, which is right and appropriate. The evidence on harmful content, addictive design and mental health impacts is serious and growing. These conversations need to happen.
What isn’t being discussed, at least not loudly, is the degree to which the current pace of AI development has been quietly subsidised by a business model that monetises children’s attention and data. The $11 billion in US teen ad revenue isn’t just profit. It’s research funding. It’s compute. It’s the next generation of models.
Regulators are, in effect, beginning to dismantle a revenue model that has cross-subsidised one of the most significant technological transitions in human history. That might be entirely the right thing to do. But we should be clear-eyed about the trade-off we’re making.
The question for 2026 and beyond isn’t just “should teenagers be on Instagram?” It’s “who pays for AI development when the current model changes, and what does the answer mean for the speed and direction of progress?”
We don’t have a clean answer. But we think it’s the right question to be asking, and we suspect it won’t stay unasked for long.
