Synthesia AI: Replacing the entire physical video production process with AI

(5 minute read)

This AI unicorn is replacing the entire physical video production process

50% of Fortune 100 companies including Microsoft and Nike use their AI product

They have 50K+ customers, up 456% year-over-year

In June 2023, they raised $90M at a $1B valuation from investors like Nvidia & Kleiner Perkins

So what’s the startup and what does it do? Here’s the story of Synthesia AI

Synthesia AI was founded in 2017 by Victor Ripabelli, Steffen Tjerrild, Matt Niessuer and Lourdes Agapito,. Theyre a bunch of AI researchers and entrepreneurs from Stanford, UCL, TUM and Cambridge.

Synthesia is an AI video creation tool that allows you to create videos without cameras, microphones or studios. It helps people create video content with AI avatars and voiceovers in 130+ languages.

It’s mainly used for teams creating training, sales enablement, customer service and marketing video content.

To use Synthesia to create video content you:

1. Select an avatar or create your own

2. Type in the text you want

3. Build your video around this text

Synthesia is like Powerpoint 2.0, it's super easy to use and it only takes a matter of minutes to generate video content. Its main aim is to replace text entirely with video.

First Phase of Synthesia

Flashback to 2017, their first product was barely functioning, took a lot of time to produce content (up to 3 days to produce a 10 second clip), and wasn’t very scalable.

Their initial customers were ad agencies and the film industry, and they did $700k in revenue their first year, but couldn’t grow at scale.

What Synthesia’s founders figured out at the time was that there’s billions of people in the world who are desperate to create video content.

If people had a way to generate video content in a 1000x more affordable and scalable way, they’d be okay with a slight drop in video quality.

Synthesia quickly built and shipped the product knowing that the quality of videos was not one-to-one with a real camera.

These videos were not a replacement for video production, but a replacement for text.

The problem: Workers are meant to read and remember entire 40+ page training manuals where the majority would forget about it

The solution: Companies can create video content in their workers’ native tongues which they would actually remember.

Using video content increases their information retention by 4-5x…

Synthesia just created a new market by unlocking new capabilities for a new set of people.

So, how do they make money?

Synthesia has 2 types of customers:

• Self-service: This is typically small businesses with little to no budget on video production. They pay a monthly subscription from $0 to $50 a month

• Enterprise: Usually used to train and enable 1000+ person workforces. These plans are customized based on the size and needs of a team

Synthesia wants to be the best in the world at creating insanely photorealistic humans that can’t be distinguished from real video. This could be scary though.

Data privacy and deep fakes

Some people expressed that technology like what Synthesia is building, cause data privacy issues - based on the data trained by their LLMs.

Synthesia does train their LLM models on clean, compliant data only. They’ve also made sure you can only produce original content on the platform.

Tech developed by Synthesia has reportedly been used to produce propaganda in Venezuela and create fake news reports in China.

Synthesia, says that it minimizes the number of these issues through asking for someone’s full, explicit consent before using their avatar.

They also do content moderation - you're only allowed to produce certain types of content using the tool. Synthesia says they’re not perfect, but always trying to improve.

Stats

In June 2023, Synthesia raised a $90M series C at a $1B valuation, bringing their total funding raised to $156.6M. Some investors they’ve gotten onboard include: Kleiner Perkins, Nvidia, Accel and GV.

Synthesia currently has over 200K users, with 55,000+ paying customers worldwide. This includes 50% of the Fortune 100…

They currently employ over 200 people, 10% of which focus solely on AI safety and ethics.

A couple of days ago, Synthesia announced their avatars could convey human emotions including happiness, sadness and frustration. 

This is a huge step towards creating ‘virtual humans’ but comes with even bigger concerns of creating fake news.