No matter how you looked at it, it was a jaw-dropping acquisition at the time.
YouTube, only about a year-and-a-half old at the time of the acquisition, had seen a meteoric rise by late 2006 – to about 72 million users worldwide.
It was deeply unprofitable and bleeding cash… and acquired by Google for $1.65 billion – a figure that was about $1 billion more than what Google’s CEO at the time thought it was worth.
It was a crazy figure. YouTube’s revenues were thought to be negligible. The company was still experimenting with how to monetize user activity on its platform. At $1.65 billion, YouTube almost certainly sold for more than 200 times annual sales.
That kind of multiple might make us shake our heads. How could that have been justified? How could Google possibly generate a return on an acquisition like that? Why would they do that?
Sometimes it helps to reframe an acquisition in a different way… to understand whether or not it makes any sense.
Instead of using a discounted cash flow analysis, or a multiple of annual sales or EBITDA, we might look at it as an advertiser might see it…
Google’s business has always been about data surveillance, collection, and monetization of that data through selling advertising.
At $1.65 billion, Google purchased YouTube for about $23 per user.
Suddenly, it feels a lot more reasonable, doesn’t it?
Generating more than $23 of advertising per user over a number of years doesn’t sound unreasonable at all.
And of course, for the $1.65 billion figure to make sense, Google’s board needed to believe in a few key future outcomes:
The other dynamic worth mentioning is that YouTube was being hotly pursued at the time – by not only Google but also Yahoo and Microsoft.
In another framework, Google felt it couldn’t afford to have any of its competitors acquire YouTube.
But regardless of the price, we all know how the deal turned out.
YouTube now has about 2.7 billion monthly active users. And last year alone, YouTube’s combined advertising and subscription revenues generated more than $50 billion for Google.
That $1.65 billion price tag looks like an absolute bargain now…
Of course, Google had to invest billions to scale YouTube to support its growth, the kind of money a tiny startup simply doesn’t have. But regardless, it was an outstanding acquisition.
Video was, and still is, the most compelling form of media on the internet and social media. Google believed it back then, and everyone understands it now… given the incredible success of Facebook, Instagram, YouTube, X, and, of course, TikTok.
But here’s the most interesting twist…
It turns out that the largest return on investment of the YouTube acquisition isn’t from advertising.
From my perspective, an even greater value from the acquisition comes as a massive video training set for Google’s generative artificial intelligence.
And ironically, this is something that the Google executives couldn’t have foreseen back in 2006.
But we can now see that value in Google DeepMind’s release last month of Veo 3, its state-of-the-art generative AI video model.
It’s ridiculous, I don’t know what else to say about it.
Of course, it can create all sorts of impressive computer graphics and animation, now with voices:
Source: Google
But most impressive are the lifelike videos incorporating audio that are so good, most people would find it difficult to tell the difference between something produced by us humans and something generated by Veo 3.
A panicked scene of a fire inside a home:
Source: Google
Reporting from a car show:
Source: Google
A standup comedy show:
Source: Google
None of the above videos are real… and they’re worth seeing in video format with the corresponding audio, unlike the GIFs pasted here.
The realism and the fidelity of what Veo 3 can produce are extraordinary. And it’s largely because of Google’s inherent advantage of owning and controlling YouTube.
And here’s the crazy part…
Creators are already starting to create short-form video just using generative AI… and posting that content on YouTube. That exponential growth in new video content will be used for training future versions of Google’s Veo generative AI model. It’s an ideal flywheel…
Google’s AI advantage with video is similar to Tesla’s advantage with autonomous technology. Tesla now has roughly 15 billion miles of data collected from real-world conditions of its cars driving themselves. No other competitor is close. No other competitor has even a small fraction of the dataset that Tesla has on self-driving.
Veo 3 may only be able to produce short videos of less than a minute right now, but it’s only a matter of time before that becomes 10 minutes, half an hour, or even a full-length movie.
Just imagine the author of a book feeding Veo their novel… and telling it to produce a two-hour movie emphasizing key parts of the storyline.
The author, with the support of some video editors and scriptwriters, could produce his own movie based on his own content. In time, no one will be able to tell if it came out of Hollywood or Veo…
But you might be surprised by what I’ll say next. I don’t think that’s where we’ll see Veo used the most…
Where I believe and expect generative AI technology like Veo to be used the most is with regard to social media and advertising.
Think about this. It’s the ultimate tool for customization and engagement…
Both creators, publishers, and social media companies can generate video specific to an individual end-viewer, because – through Google – they can deeply understand who that person is. After all, they have years of data surveillance at their fingertips, again, thanks to Google.
So in fact, using agentic AI – and incorporating something like Veo – content creators will be able to engage individual end viewers with content, analyze engagement metrics and conversion rates almost instantaneously, and even take into account facial expressions and responses by the viewer (remember, every smartphone has a camera on the front for a reason…).
If the agentic AI isn’t getting the desired response from the viewer, it will iterate and improve the video – in real time – until it does.
The agentic AI tracks the viewer’s eyes moving adrift – it optimizes for shock effect to re-engage…
The agentic AI picks up a smirk – the viewer liked that part of the video… more of that.
And it’s not just about advertising… obviously bad actors, political parties, and nation states will use this technology to infiltrate, manipulate, and conduct psychological operations on populations around the world, too.
We already got a taste of what that feels like during the pandemic, when false information was propagated, and evidence-based information was censored.
But this is a completely different level.
It will be nearly impossible to tell fiction from reality.
Which leaves us with two realities…
Media platforms will have to document and demonstrate the validity of their content, and earn the trust of their users…
Or individuals will need to demonstrate critical thinking with every piece of content they consume.
And because of the level of customization and personalization of video content, it will only become that much more addictive for the average user.
Jeff
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
The Bleeding Edge is the only free newsletter that delivers daily insights and information from the high-tech world as well as topics and trends relevant to investments.
