(#96) 🇩🇪 German auto tragedy in China; Tesla has a software challenge
How a Bayer CFO is thinking about AI ROI
Today’s Insights:
The Nobel Prizes in Physics and Chemistry go to AI
SpaceX is the new dream for humanity
German Auto Tragedy in China
From the PLUS version (paid):
How a Bayer CFO is thinking about AI ROI
What went wrong with Starbucks’s mobile ordering strategy
The curse of the Michelin star
On to the update:
Strategy
Nobels for AI
Geoffrey E. Hinton (from the University of Toronto, Canada) won the Nobel in physics “for foundational discoveries and inventions that enable machine learning with artificial neural networks”.
Demis Hassabis (co-founder of Google DeepMind) & John Jumper (DeepMind Research Director) were awarded the Nobel Prize in Chemistry for “their work developing AlphaFold, a groundbreaking AI system that predicts the 3D structure of proteins from their amino acid sequences”.
From DeepMind’s press release:
“[...] Before AlphaFold, predicting the structure of a protein was a complex and time-consuming process.
AlphaFold’s predictions, made freely available through the AlphaFold Protein Structure Database, have given more than 2 million scientists and researchers from 190 countries a powerful tool for making new discoveries. The AlphaFold 2 paper, published in 2021, remains one of the most-cited publications of all time.
AlphaFold’s contributions to science have been widely praised, and among its recognitions are the 2023 Albert Lasker Basic Medical Research Award, the 2023
Breakthrough Prize in Life Sciences, the 2023 Canada Gairdner International Award, the 2024 Clarivate Citation Laureate award, and the 2024 Keio Medical Science Prize Award. [...]
Congrats and well done! Physics. DeepMind
SpaceX is the new dream for humanity; some thoughts on Tesla’s “We, Robot” event
Here is Elon Musk on X.
Now I really believe it can be done. Europeans lost the space race… in 2013. Elon Musk, Starship, Europeans
Last week Tesla announced its first EV self-driving car - Cybercab - together with Robovan.
Available in 2026…maybe.
As with all prior announcements, Tesla shares went down on the NYSE by 6%. Here the public is split between Tesla (which has a Level 2 autonomy - out of 5) and Waymo (which has almost Level 5).
Here are some ideas worth mentioning:
1/ A Waymo car has over 50 cameras, sensors, Lidar technology and costs around 200k USD. It’s hard to scale this business. Waymo has a hardware problem and an approach to self-driving that will never reach Level 5
2/ Tesla is going to use only vision and AI technology to solve this problem. They are not there yet, but once they are the scaling part will be the easy thing to do.
So, I’m super bullish on Tesla. LINK
German Auto Tragedy in China. LINK
Maybe the best conclusion is the quote from the article:
“Zhou, an IT engineer living in Wuhan, had to endure the frustration after buying an ID.4 in early 2022. With screens going black several times in the middle of driving, he got glitches instead of German quality. Updates were also regularly behind schedule or available only via the dealer.
He’s now seeking a replacement. It will be another EV, but this time “I won’t visit any German dealerships,” he said. “I’ll only go for local brands, or Tesla.”
Here is what Europe needs to do:
1/ Develop its software on open source. When not possible, partner with American companies (e.g. VW with Rivian)
2/ Integrate vertically
3/ Develop battery technology
4/ Move away from relying on Chinese tech and raw materials.
5/ Time to understand that consumers have less money. Chinese consumers now prioritize technological innovation, digital features, and affordability over traditional selling points like horsepower and brand prestige. This will become a global trend.
Artificial Intelligence
Generative AI 2nd phase
Interesting article from Sequoia Capital on the development of Gen AI. There are multiple takes, but at least these three:
1/ Generative AI is evolving from relying solely on pre-trained, rapid responses (System 1 thinking) to integrating more thoughtful, deliberate problem-solving (System 2 thinking). This shift allows AI to handle more complex tasks that require reasoning and inference at runtime, such as coding or mathematics. This mirrors advancements in AI, such as AlphaGo’s victory over Lee Sedol in Go, where AI surpassed human capabilities by leveraging inference-based reasoning. Future AI systems will prioritize this deliberate, real-time reasoning to address more nuanced and novel tasks.
2/ A new cohort of agentic AI applications is emerging, characterized by their ability to perform tasks traditionally reserved for humans. These applications, such as AI software engineers, medical scribes, and penetration testers, lower the cost of services by automating tasks that previously required significant human labor. This trend points to a significant shift in the economic landscape, where AI is not just replacing software functions but is also automating entire categories of human work, expanding markets, and creating new opportunities for AI-driven services.
3/ One of the key advancements in AI is the concept of inference-time computing, where AI systems dynamically scale computational power based on task complexity, leading to better reasoning capabilities. This shift from pre-training to scalable inference opens new possibilities for tackling long-term problems, such as scientific discoveries or unsolved mathematical theorems. The introduction of new scaling laws for inference-time computing suggests that the future of AI will focus on reasoning-driven architectures that continue to improve as more computational resources are allocated during the inference phase.
It’s time to continue building in this space. LINK
AI Agents are the future of white-collar companies
Everything that means working from home will be eventually automated with AI Agents.
Here is The Verge with many insights, but I chose only the following:
1/ AI companies, including OpenAI, Google, and Microsoft, are heavily investing in developing AI agents - autonomous programs designed to perform tasks with minimal human input. These agents are seen as a progression from traditional bots (ie. dummy robots = RPA), capable of interacting with environments, learning from feedback, and making decisions. They represent a shift in how companies envision future automation, aiming to revolutionize productivity by handling complex tasks such as scheduling, customer service, and even legal or tax services.
2/ Despite the technological potential, the rush to develop and deploy AI agents is driven largely by market pressures (ie. cost cuttings, etc). Companies like OpenAI and Anthropic are sitting on expensive AI models and are eager to find practical applications that can justify their costs. As a result, there's a strong push to monetize these agents, with venture capital flowing into startups that promise to leverage these autonomous systems for businesses and consumers. However, these agents remain costly to operate due to their need for substantial computational power during reasoning tasks. Salesforce wants to charge you 2 USD per task accomplished. That’s too much, I’d say, but depending on the difficulty of the task it might be overall much cheaper with AI.
3/ While AI agents hold promise, they currently encounter practical limitations. Issues like handling multi-step workflows and unexpected scenarios demonstrate that agents are not yet significantly more useful than simpler bots or assistants for everyday tasks. Moreover, concerns about trust, especially in high-stakes environments like law or finance, and the potential for misuse (e.g., spam or deception) and others. The technology is still evolving, and further advancements in reasoning and computational efficiency are needed before AI agents can deliver their full potential.
Things Happen
Export your brain with AI. LINK
How FIFA was outplayed by Electronic Arts. LINK
China’s robotic baristas serve coffee 90% cheaper than traditional coffee shops. LINK
Can the French make good wine in California? LINK
From beans to chips, vertical integration differs from older models. LINK
Toblerone could soon become luxury goods. LINK
Industry reports
Energy consumption
The International Energy Agency (IEA) predicts that air conditioning will significantly drive global electricity demand. As temperatures rise and incomes increase, demand for electricity from home cooling systems could surge by 280% by 2050. FT, IEA
Poland overtakes Russia in the value of exports for the first time. LINK
On the fertility front: Europe is disappearing. LINK
Maybe the response to this ongoing crisis might be in Georgia or…Mongolia.
State of Fintech Q3’24 Report, from CB Insights. LINK
Curiosity corner
The Beatles Effect. LINK
[Study] Christopher Columbus was a Sephardic Jew from Western Europe. LINK
Trabant quality control. LINK
AI-accelerated Nazca survey nearly doubles the number of known figurative geoglyphs and sheds light on their purpose. LINK
Thank you for being an onStrategy reader!
Interesting newsletters 💡
This newsletter continues below with additional strategy insights exclusive to premium subscribers. To unlock them, two annual presentations, and more, simply click the button below. 👇
Keep reading with a 7-day free trial
Subscribe to onStrategy to keep reading this post and get 7 days of free access to the full post archives.