20VC Newsletter - 2nd February 2025
Here are the transcripts and top takeaways from 20VC episodes released this last week.
Monday’s episode with Wayne Ting, CEO @ Lime:
Download the full transcript:
My 7 key takeaways:
Everyday 3% of Lime Scooters Break
The industry used personal scooters that weren’t designed for commercial use.
The decay rate was 3% – the average scooter only lasted 30 days.
In the course of 30 days your entire fleet was gone.
How a Change of Batteries Changed the Business of Lime!
We used to pick up every scooter & bike in a city, drive back to a warehouse & charge them.
We switched from fixed batteries to swappable batteries.
We could replace hundreds of batteries with one cargo – it improved productivity & reduced the cost per swap.
How Machine Learning Tells Lime Where to Put Bikes
We have a machine-learning algorithm that predicts demand in cities.
The algorithm also estimates potential revenue in each block.
If the potential revenue of moving a scooter/bike to another location is higher than the cost of moving it, it will create a task for our movers.
Not Firing More People Was the Worst Mistake Made at Lime
We did 2 rounds of layoffs within 3 months.
Not cutting deep enough was one of my worst decisions we’ve made.
If you have to do layoffs, do it once & do it hard.
How Uber Came to Invest $140M Into Lime
We lost 90% of our revenue during COVID & had 20 days of runway left.
I pitched Dara to sell Jump to us & merge the businesses.
Uber invested $140M & we became their micro-mobility option.
Why Hype Cycles Are Bad for Companies
It allows them to avoid learning hard lessons.
It allows them to not focus on the things that truly matter.
If they can always raise money it masks their bad operations.
Stop Worrying About a BS Valuation and Go Public
Too many companies are unwilling to go public until they hit a valuation they had in 2020.
That valuation is fake; you are making irrational business decisions because you are trying to hold onto a pipe dream.
They should face reality as soon as possible.
Wednesday’s episode with Jonathan Ross, Founder & CEO @ Groq:
Download the full transcript:
My 6 key takeaways:
How DeepSeek Got as Good as OpenAI in 6 Months
They beat the scaling law problem by training a model, generating data & retraining with it.
They took a shortcut & used a preexisting model.
It’s true that they spent $6M but they spent even more scraping from OpenAI.
Is DeepSeek a Tool Used by the CCP to Control the US
A lot of answers are already heavily filtered.
You cannot ask about topics like Tiananmen square.
BUT it can give a detailed explanation of why TikTok should not be banned in the US.
Why $500BN For Stargate is Still Not Enough
Putting a model into production is always super expensive.
When I was at Google we always spend 10-20x more on inference than compute
Now the models are being given away for free
Why Open Source is the Secret to DeepSeek Winning
If DeepSeek wasn’t open source it wouldn’t have gotten the excitement it has today.
Open source always wins.
Linux won back when people didn’t trust open source.
OpenAI Has to Open Source Now
If I was in Sam’s position I would be gearing up to open source my models in response.
If they are going to lose, they might as well open source & win the users.
I do not understand why they’ll still try and keep it internal.
Should the US Use Tariffs on China and Lessons from the Cold War
We need to have mutually assured destruction.
There needs to be a sort of automated response.
If they subsidize an industry, we will automatically subsidize the equivalent.
Friday’s episode with TS Anil, CEO @ Monzo:
Download the full transcript:
My 5 key takeaways:
Why No One Has Won Us Fintech Markets
There is too much cheap capital chasing US fintech.
This creates a context where most fintech entrepreneurs are incentivized to grow their user base.
We have too many mediocre products fueled by CAC dollars that don't truly resonate with users.
How Does Monzo Actually Make Money
Three ways:
Transaction-based: Interchange, FX...
On-balance-sheet lending: Loan, overdraft, Monzo Flex
"Good fees": Subscriptions, marketplace & mortgage origination, interest on savings products
They are all around ⅓ each.
Why Monzo Will Not Do Mortgage Financing and What Part of Mortgages Are a Good Business
The problem we could solve is the front end of the customer journey.
There are enough large balance sheets in the country that would happily distribute mortgages through us.
We would rather not deploy our capital to fund the mortgage asset itself.
What You Need to Lead a Large Company
Three key areas:
Setting company ambition and product vision
Operating close to the product, users, and value creation
Scaling execution by building the right leadership, team, and culture
It’s like a triathlon – You need to be great at all 3.
When Will Monzo Go Public?
We have the luxury of time.
We have supportive investors & plenty of capital to build out our ambition.
We’ll make a great public company one day but we’ll get there when we get there.
Let us know what your big takeaways from this week’s shows were in the comments below!
Thank you for reading, and don’t miss the great guests we have next week:
Nabeel Hyatt, GP @ Spark Capital
Max Levchin, Co-Founder & CEO @ Affirm
Brian Tolkin, Head of Product @ Opendoor
Thank you for reading 20VC.
This post is public so feel free to share it.
Thanks for reading 20VC! Subscribe for free to receive new posts and support my work.
Why have i just found this now?! So many nuggets in 20VC podcasts that you can easily miss them. Allowing me to download the transcript and interrogate it with Notebook makes me happy.