(#32) Netflix’s ad model is working; Tesla can be 10x; META introduces Llama and 8 habits that could add up to 24 years to your life
(#32) Netflix’s ad model is working; Tesla can be 10x; META introduces Llama and 8 habits that could add up to 24 years to your life
Most (all?) companies present in China should have a decoupling strategy
Strategy
Derisking
One of the main themes of the first half of the year was the decoupling of foreign companies from China. The official term is “derisking”. Main destinations: S-E Asia (Vietnam, Thailand, India), Eastern Europe, and USA. Now, HP announces that it moves production. LINK
What will happen next can be summarized by the quote:
“How did you go bankrupt?"
Two ways. Gradually, then suddenly.”
Netflix
The company has reported gaining 5.9m new subscribers, mainly as a result of password sharing crackdown. Here is a comment of mine from the 27th of April:
“Can Netflix be free?
[...]
4/ More people, more money. So, the ads system on Netflix runs the same as the other Social Media platforms: companies (advertisers) are bidding for the ad space. (4 minutes/hour). So, the more people are on this ad tier, the more expensive becomes the ad, and the more revenue per user for Netflix.
So, answering the initial question: Yes, of course! Not only that the company will make more money from ads, but they can decrease the price or even make it zero” - LINK
Today:
Key take 1: Nothing has changed, except that in the USA there is only one ad plan and two more expensive ad-free plans.
Key takeaway 2: the company will have more free cash flow because some productions will not start/finish due to the writers and actors' ongoing strike in the USA. Solution? Produce more content on other continents.
Tesla
2023 can be the company’s year of consolidation, for several reasons:
1/ Most Gigafactories are up and running: Texas, Berlin, and Shanghai
2/ Model 3 and Y are in the top three cars sold in most developed countries (especially in the EV sector); Model Y was the most-sold car globally in Q1, surpassing Toyota Corolla and VW Golf.
3/ Q2 2023 brought $25bn in revenue for the company
4/ Tesla's full self-driving will be 10 times safer than the average human driver, still the full autonomy is not there yet. Having so much data from all the cars sold give them an unparalleled advantage over other companies.
5/ The Supercharge network has reached 50,000 connectors in 5,0000 locations.
3/ The much-expected Cybertrack is expected to launch this year. (observation: there are 1.7m orders for the car so far)
4/ New Gigafactories in the sight: Mexic and India
What surprised me in the investor’s call was Elon saying: “I really see a path to 10x (in valuation)”. LINK
Meanwhile, Tesla’s gross margin is the same as Toyota's and 5 pp. below Mercedes
TikTok goes after Twitter & alike
The company has recently announced that you can express your creativity with text too.
Some takeaways from the announcement:
1/ Text is such a weak medium compared to video/photo. That’s why TikTok reached +1bn users so fast and Twitter still struggles with 450 monthly active users (MAUs) after +15 years of life
2/ If there is a chance that a feature can bring you 10-20m people using it, why not? Especially with the need to post on other platforms diminishing. LINK
Twitter becomes X
The ongoing mess can be summarized by two things:
1/ Product people can lead software companies (and vice-versa)
2/ The company had a reputation for not delivering on anything for +10 years and the reputation precedes anything that happens even today.
So, there is no surprise that the new name is not registered by the company or that the company will not become a Super App like WeChat because it doesn’t have the conditions and the laws in favor (ie. competition laws). Others have tried and failed.
Matt Levine from Bloomberg asks why Elon actually bought Twitter:
“I guess my question is, what was he paying for? Musk didn’t want Twitter for its employees (whom he fired) or its code (which he trashes regularly) or its brand (which he abandoned) or its most dedicated users (whom he is working to drive away); he just wanted an entirely different Twitter-like service. Surely he could have built that for less than $44 billion? Mark Zuckerberg did!” (LINK)
Given how much Elas has spent (ie. around 10% of his net worth) and how is actually using and managing the platform it seems that he did it mainly for fun. I hope to be wrong. LINK
Most banks (all?) don’t know software
Funny story on how Deutsche Bank AG bought Postbank, another German bank, in 2010 and tried for 13 years to integrate the systems. Crazy! LINK
“Inside one of the lender’s buildings in Frankfurt, a 200-strong team of retail bankers and IT specialists were racing to complete the final 2,833 tasks needed to transfer the remaining data of German lender Postbank’s 12mn customers over to Deutsche’s computer systems.
As the meticulous plan, rehearsed several times in previous months, unfolded, an unsuspecting member of the team opened a dishwasher in mid-cycle, with the escaping steam setting off a smoke detector and forcing the evacuation of the building. Precious minutes were squandered before firefighters gave the all-clear.”
Stripe & Revolut will eventually rule over the banking system.
Knowledge Partner: INDE MBA
Are you exploring starting an MBA? Participate on campus to get a sneak peek at some classes on the 13th of September.
The program has a double certification (ASE Bucharest and CNAM Paris) and is AMBA Accredited. Register for the sneak peek HERE.
More information about the program: https://inde.ro
Artificial Intelligence
Meta introduces a new LLM
This announcement seems to have passed largely unnoticed. When it comes to generative AI (LLMs) there are two main costs:
1/ Training costs
2/ Inference costs (calculating the answer, which btw is probabilistic, not deterministic…hence costs muuuch more)
Ok, the model is weaker than GPT-4, but it has the following advantages:
1/ It can be downloaded locally and adapt it
2/ Privacy
3/ Once updated any model can be scaled in the cloud (Microsoft?) and pay accordingly.
In the end, Meta tries the open strategy approach with the LLM as Stabble diffusion did 9 months ago. However, there are two observations:
1/ The model is free only for companies with less than 700 million users (hint! Snap, Telegram, and maybe Apple); above this number, you need a license from Meta
2/ If run locally Apple will be one of the main winners due to their powerful chips. LINK
A great article from Financial Times (no paywall) on how Google scientists pioneered the current Ai revolution. LINK
Data
McKinsey Technology Trends Outlook 2023
Top 3 tech trends
1/ Applied AI
2/ Industrializing Machine Learning
3/ Generative AI
….and 12 more. LINK
'Barbie' reportedly had a $150 million marketing budget — more than the movie's actual budget. LINK
Bolivia has now the world's largest lithium reserves. LINK
Just like with CO2 emissions, the most important polluters are not the USA or Europe. LINK
Five Big Ideas Shaping Tomorrow:
1/ How Factories in Space Could Solve Problems on Earth
2/ Why Touch is VR’s Missing Ingredient
3/ Harnessing the Ocean Could Help Fight Climate Change
4/ Generative AI Will Transform Medicine
5/ Will India be the Next Semiconductor Superpower? LINK
Student debt
Nearly 1 in 3 student borrowers say they spent money they didn’t think they’d have to pay back. LINK
Skift Research’s State of Travel 2023 Report. LINK
Top contributors to this year's S&P 500. LINK
Outside Interest
The pandemic meant broken habits. Americans are becoming a nation of early birds. LINK
8 habits could add up to 24 years to your life. We know them all, but a recap is always good. LINK
Reverse dumbing-down: can you teach an old dog new tricks? An insightful analysis by Alexandra Thompson. LINK
Barcelona has more dogs than kids (0-12 years). LINK
The ongoing heat wave is affecting the whole of Europe, especially the south.
☕️ Did you enjoy the newsletter? Buy me a coffee ✅