AI Is Finally Getting "There"

Open AI's GPT4o is getting astounding, and we are rapidly approaching the future.

About six months ago, an old friend came to visit me. It was around the release of GPT-4, or maybe GPT-4 Turbo, and he had just started experimenting with it. He told me how great it was, how it was now his go-to source for information, and how he had set the "action button" on his iPhone to bring up ChatGPT immediately.

"Cool, bro," I said, "that's really cool!"

He's my friend, so I didn't want to rain on his parade. But having done some serious deep dives into the state of AI and large language models (LLMs), I knew his faith in it was a bit misguided. When he, while visiting, asked it, "What bar has the best whiskey selection in Las Vegas?" its recommendations didn't have a ton of value. It more or less suggested the top-rated bars it knew of that had "whiskey" mentioned in a review.

Since the release of GPT-3 in 2020, I've followed OpenAI's progress closely. The potential was obvious right away. "In 10 years," I thought then, "this thing will be writing novels and running my business."

Fast forward to a few days ago with the release of GPT-4o. It's not quite ready to write a novel or run my business, but I'm finally on the same page as my friend. It is now my go-to source for all kinds of information, and more interestingly, quick analysis.

And what’s even more amazing is that OpenAI didn’t think it was enough of a step forward to call it GPT-5!

When I wrote my last article, I didn’t bother to look up the CPI stats myself; I simply asked GPT-4o to find them and present them in my preferred format. The other day, when I wanted to check the statistical probability of an event, I had it simulate the situation for me in Python. Both of those requests took less than 30 seconds.

GPT-4o is finally there, and that is going to have massive consequences (mostly good) for equity markets and life as we know it.

AI Is Goliath, Not David

The first big misconception I see among finance people who are not that tech-savvy is the idea that AI is going to “disrupt” the tech giants. A guy who I enjoy reading quite a bit even said recently that he was short Google. I think this is a huge, huge mistake, and I think this for a few reasons:

1. Training an LLM takes a large amount of data. The big tech giants have been storing the most high-quality data of any companies around. They’ve been storing it, and they are profiting on it. If there is any edge to be had in data, they will have it. Caveat: Sam Altman recently mentioned on the All In Podcast that he didn’t think data was going to be the big difference maker for LLM models. We will see.

2. Right now there is an arms race happening in AI research and development. There are no companies on earth better positioned to win this arms race than the tech giants like Google, Microsoft, Apple, Meta, and Amazon. They can cavalierly shovel billions into this. And it takes billions. On META 0.00%↑’s most recent earnings call Zuck borderline bragged that the company’s cap ex costs would be rising from $37b in 2023 to $40b in 2024. A lot of that spending is going to buying AI infrastructure.

3. The tech giants have the users. In his book Zero To One, legendary investor and contrarian thinker Peter Thiel argued that a proprietary technology must be at least 10 times better than its closest substitute to create a sustainable competitive advantage and get users to switch. There is no way an AI startup (like OpenAI) beats the incumbents that badly, and if they do…

4. Tech giants will buy the winners before they win.

Take the case of OpenAI. It’s unclear if they will build a defensible moat, but they are already in bed with Microsoft in exchange for money and compute power. In return, Microsoft secured the rights to use OpenAI’s tech in their products. That deal might go down in history alongside the YouTube and Instagram acquisitions as one of the best deals ever.

The tech giants will be the winners and betting against them is insane.

What Does it Mean to Win in AI?

AI will infiltrate every piece of software, transforming all SaaS tools into sophisticated “wrappers” for the most advanced language models. Who will they pay for this privilege? The tech giants.

More incredibly, LLMs will soon evolve into “virtual employees” performing vast amounts of work on our behalf. Forget Elon’s robots; soon, Google will license you access to the world’s best programmer: an LLM residing in Google's cloud.

I cannot overstate the value these technologies will create. All knowledge work, sooner than you might imagine, will be rented on demand from cloud providers.

I’m going to give you a long list of all the jobs that will get replaced by these LLMs. You don’t have to read every one, and I’m sure I’m missing a bunch, but just take a look.

Administrative and Clerical Work

1. Data Entry and Management: Automating the input, processing, and management of data in various formats.

2. Scheduling and Calendar Management: Coordinating meetings, appointments, and reminders.

3. Email Management: Drafting, sorting, and responding to emails.

4. Document Creation and Formatting: Generating reports, presentations, and other documents.

5. Customer Service: Handling inquiries, complaints, and providing information through chatbots and virtual assistants.

Financial Services

6. Bookkeeping and Accounting: Automating financial transactions, reconciliations, and reporting.

7. Tax Preparation: Completing tax forms, identifying deductions, and ensuring compliance.

8. Financial Analysis and Reporting: Analyzing financial data, forecasting, and generating insights.

9. Investment Management: Portfolio analysis, algorithmic trading, and personalized investment advice.

Legal Services

10. Contract Review and Drafting: Analyzing legal documents, drafting contracts, and identifying risks.

11. Legal Research: Searching for case law, statutes, and legal precedents.

12. Compliance Monitoring: Ensuring adherence to regulatory requirements and standards.

Human Resources

13. Recruitment and Onboarding: Screening resumes, scheduling interviews, and managing onboarding processes.

14. Employee Training and Development: Creating training materials and personalized learning plans.

15. Performance Management: Monitoring and analyzing employee performance data.

Marketing and Sales

16. Content Creation: Writing articles, blog posts, and social media content.

17. Market Research: Analyzing market trends, consumer behavior, and competitive analysis.

18. Sales Automation: Managing leads, CRM updates, and customer follow-ups.

19. SEO and Digital Marketing: Optimizing content for search engines and managing online advertising campaigns.

Healthcare

20. Medical Documentation: Transcribing and managing patient records.

21. Diagnosis and Treatment Planning: Assisting in diagnostic processes and suggesting treatment plans.

22. Telemedicine: Providing remote consultations and follow-ups.

Education and Training

23. Tutoring and Coaching: Providing personalized learning support and mentoring.

24. Curriculum Development: Designing and updating educational materials.

25. Grading and Assessment: Automating the evaluation of assignments and tests.

Research and Development

26. Data Analysis: Processing and analyzing large datasets for insights.

27. Literature Review: Summarizing and synthesizing research papers and articles.

28. Experiment Design: Assisting in the design and simulation of experiments.

Creative Industries

29. Graphic Design: Creating visual content, logos, and marketing materials.

30. Video and Audio Editing: Automating editing processes and content creation.

31. Game Development: Designing game mechanics, narratives, and AI behavior.

IT and Software Development

32. Code Generation: Writing and debugging code based on specifications.

33. System Monitoring and Maintenance: Automating system health checks and updates.

34. Cybersecurity: Detecting and responding to security threats.

Professional Consulting

35. Business Strategy: Analyzing business data and providing strategic recommendations.

36. Risk Management: Identifying and mitigating business risks.

37. Supply Chain Optimization: Improving logistics and supply chain efficiency.

Sooner rather than later Google and the others will either rent you those services directly or rent their LLM to a company that has customized it to provide that service to you.

Take a second and really think about what a world like that looks like, and how much value accrues to the tech giants. Having trouble? Here, I’ll help you with a few rough estimates of how much market cap is going to get slurped up. (Who do you think helped me dig up these numbers?)

• Legal Services: $700 Billion

• Human Resources: $500 Billion

• Marketing and Sales: $600 Billion

• Financial Services: $20 Trillion

• Education and Training: $10 Trillion

• Research and Development: $2.4 Trillion

• IT and Software Development: $5 Trillion

• Consulting: $250 Billion

Now, just to be clear, I’m not saying the tech giants will totally eat those entire industries. But they will either completely dominate them and offer them as an LLM, or they will stick their greedy hand in and get their $99/month from any business that wants to operate in them.

Trading This

This post is getting long, and has been fairly wide in its scope, so I want to bring it back down to earth. How might one trade this?

Well, hopefully it’s obvious but I’m buying the big tech stocks. I think it’s clear AI is here, it’s clear LLMs are the way AI will manifest, and the tech giants will be the big winners.

“But,” you might say, “the tech giants are just levered beta. Is that really the best you’ve got? Give me the Alpha!”

Well… No, that’s not the best I’ve got.

DUOL 0.00%↑ got some negative press after the GPT-4o model was unveiled for going down a few %. It’s now down about 12% on the week.

Duolingo and many, many stocks trading on the Nasdaq will get their lunch eaten by this tech.

So the real move would be to long the tech giants, short the dusty Duolingo-style tech stocks, and watch as the wealth inequality gets worse and worse while people are replaced by AI and the world is sorted into “people who bought the big tech stocks” and “the people who live in the ghetto we don’t go to”

Hyperbole of course, but directionally correct. But that’s a post for another day.

---

Please also always remember... none of this is financial advice. I’m not a professional. I quite literally don’t know what I’m doing. I’m just a guy, with a keyboard, who scored high enough on some standardized tests to think I can beat the market.