We are so used to video streaming services like YouTube. More and more cable customers decide to “cut the cord” and just switch to video streaming. As a side note – I was surprised to see that rooftop TV antennas were not popular in the US (recent survey data shows that only 19% of US homes have TV antennas).
Now imagine the world where you could pick any book (and
come up with your own story), choose actors you like (real or fictional) and
watch a movie that is generated for you on the fly (and personalized based on
your other preferences).
This might happen sooner than we think. Let’s extrapolate
based on some known data points.
Napster (the pioneering peer-to-peer music file-sharing
service) was launched in 1999. YouTube was founded on February 14, 2005 with the
first video uploaded in April 2005.
We see a 6-year gap between the audio streaming and video
streaming product launches. This delay wasn't just about tech - it involved Internet
connection/bandwidth improvements, legal battles (new unchartered territory for
the music industry, piracy concerns, etc.), and cultural shifts (from
downloading files to streaming user uploads). Video required more data and compute
processing power than audio.
Side note: Spotify was created in 2006 and launched in 2008 –
arguably the first really popular and usable music streaming service. Typically,
it takes several attempts and iterations to launch a service that wins customers’
hearts.
Estimating the Launch of an AI-powered On-Demand Movie
Generation Service
I was really impressed by Suno AI when I first tried it.
Anybody can generate a professional grade song with a single prompt. It can use
your own lyrics and generate music in the requested style, or it can even
generate lyrics for you. This is literally one step away from an AI-powered radio
station streaming generated music hits 24/7.
Suno AI became widely available on December 20, 2023 – so let’s
assume 2024 as the launch date. To me Suno represents a breakthrough in
AI-generated music similar to what Napster did to audio streaming. So, when
might we see a "YouTube-like"
service for generating full movies on demand?
Yes, I’ve seen Sora 2 and what ByteDance does in this space
these days is truly impressive. But I am talking about an easy-to-use service
where a simple prompt from a user generates a full-size movie based on that
request and customer preferences.
If we apply the same ~6-year timeline directly, a full-fledged AI service of this kind might emerge around 2029-2030. However, Moore's Law-like advances in AI models might change these timelines slightly.
Factors Influencing the Timeline
- Technological Hurdles: Current models handle short / single digit (1-8) minute clips well, but full movies still need breakthroughs in long-form coherence and consistency. It needs to get to at least 60-90 minutes to make it viable. As a precursor to this I am expecting to see somebody combining LLMs for scripting with video generation (Suno can do this for music already).
- Legal and Ethical Issues: Napster had a fair share of legal/copyright battles. AI content generation already faces lawsuits (e.g., over training data). Hollywood unions fear job losses. Regulations (e.g., EU AI Act) might delay consumer launches.
- Acceleration:
Unlike the 2000s, AI benefits from cloud compute, open-source models, and
massive investments. We went from ChatGPT to Sora in approximately just ~2
years.
My Estimation
Based on the Napster-YouTube dynamic / analogy but adjusted
for AI's faster evolution:
- Optimistic:
A beta (first gen) service could launch by early 2027, focusing on
short films (~30 minutes).
- Realistic:
Widespread, usable "on-demand movie" platforms (i.e. "Suno
for films") by mid-2028, once models handle 90-minute outputs
reliably.
- Conservative:
If legal hurdles slow progress, closer to 2030, mirroring the full
6-year gap.
