Nerds,
I salute you from Miami, as I continue my summer wedding tour - this time in “La Isla del Encanto” - San Juan, Puerto Rico.
This week we had the first Investors Club Masterclass - on how to evaluate AI companies.
12 goodies worth sharing
1. You can map out every AI company in 5 layers:
Apps: From ChatGPT to Adobe - any application leveraging AI in their features.
Models: The reasoning models powering your apps - GPT 5.2, Sonnet 5, NanoBanana, etc
Data: The information we input into the models for their pattern matching. This can be public, licensed, proprietary or synthetic.
Cloud compute: The gateway granting the apps and models access to the hardware
Hardware & infrastructure: The physical layer - data centers, chips, computers with crazy processing power.
2. OpenAI has a billion users, millions of paying subscribers, and is almost broke. Usage-based cost has outpaced subscription revenue and most applications using AI aren’t using OpenAI models. OpenAI might have won consumer, but behind your favorite apps, you’re using Anthropic, Deepseek, GLM, Google, and others.
3. ChatGPT defined the AI consumer boom, but the term was coined in 1956. Development in AI started in the 1940s. AI winning at chess & Jeopardy existed in the early 2000s. But ChatGPT reached 1M users in 5 days. Almost 1B in less than 3 yrs.
4. Tokens are the atomic unit of AI models. You can imagine them as 3 to 4 characters of text. The model predicts the next one, over and over - generating responses to your demands. It’s our new pricing & productivity measurement unit.
5. Data continues to be the moat. Software is easier than ever to make. Models’ token pricing has dropped 90% in 2 years. Find a way to generate proprietary data and buyers (i.e. their AI agents lol) will come find you.
6. Your portfolio is already more exposed to AI than you realize. ~40% of the S&P 500's value sits in 10 companies. 9 of them are AI. If you've ever bought an index fund, you already have AI exposure, even without SpaceX or NVIDIA in your portfolio.
7. Just because it’s good tech doesn’t mean it’s a good investment. IBM built an AI that beat Jeopardy in 2011. The tech was awesome, but didn’t get 1 million users in 5 days like ChatGPT. In a world where software gets commoditized, distribution is queen.
8. You can capture AI value without creating it. Apple owns none of the AI stack - no model, no cloud, no data centers, and still runs a 79% margin on it, leveraging their distribution to gain access to OpenAI models at a small price. Master distribution first, then win on partnerships.
9. There is still much more AI value to be created.
We’re in the ANI phase of AI - “Artificial Narrow Intelligence”. This means AI can do specific tasks really well.
Up next, we’ll reach AGI - “Artificial General Intelligence” - when AI is as flexible in their reasoning as a human about any topic.
Then, ASI - “Artificial Superintelligence” - the “apocalyptic AI” that is better than a human. Think Singularity moment, AI supreme world leaders, etc..
10. AGI won't announce itself. Capabilities are doubling roughly every 6 months, but there’s no single test that’ll say "hey, we made it!" It’ll happen gradually, then suddenly.
11. The companies mastering AI are hiring more, not less. Data is showing more AI usage means more productivity, more ideas, more ambition, and ambition needs headcount. The jobs are shifting, not disappearing.
12. There’s 5 possible scenarios for the AI endgame, and only one of them doesn't involve a fight.
AI takes over and generates value with no humans involved
Countries (i.e. US and China) go to war over control of the tech
Countries actually collaborate on shared regulation (imagine?)
Big companies take over, dominating while governments just referees
The tech plateaus and never reaches its potential
And like good wine, we paired each of these with stocks to look out for.
If you want to join us as we grow our portfolios - I’m opening the next (and last!) cohort of Wealthy Women Don’t Wait.
Enjoy an early bird discount until this Sunday, July 12th.
Agree or disagree? Let’s gossip! Reply and would love to chat.
See you in cyberspace,
Jules 🤸🏻
