Khanmigo AI, Public vs Private

Khanmigo AI Example

Many of you are already familiar with the online education resource Khan Academy. It’s one of my favorite educational organizations. In the mid-2000s, Salman Khan’s YouTube videos on math education went viral. The Gates Foundation asked him to focus full-time on developing his content into an accessible, high-quality online education platform as a philanthropic endeavor. I remember this well because I graduated with a math degree in 2010 when Wikipedia and Wolfram Alpha were the only free online resources, and textbooks cost $250-500 each. Today, Khan Academy has over 100m registered users globally. They recently launched Khanmigo, which integrates AI as a personalized tutor. You may have seen this in the OpenAI demo day video last week.

This is interesting for a few reasons. One is that Khan Academy did not have to invent LLMs and machine learning to roll this feature out to its global user base. The other is that this is a perfect example of the deflationary aspect of technology.

Khanmigo can provide personalized conversational AI-based tutors on demand for $4/mo for students and free for teachers. The average cost of a tutor in New England is around $40/hr. I like to think of AI as supplementary rather than as a replacement. At this price point, every student can afford unlimited access to a world-class tutor in any language at anytime. It’s a far larger market than in-person tutors could ever supply.

It’s hard to estimate where AI’s value will accrue from an investment perspective. LLMs are incredibly flexible and, in my view, similar to the invention of the spreadsheet. If you went back in time and asked where the value of the spreadsheet would accrue, it turns out it pretty much helped everyone and became fundamental to modern business and organization. Check out this Microsoft Excel commercial from 1992. I think that’s probably where these AI tools are heading, and this Khanmigo example is one of many positive general outcomes that have emerged from the rollout of this invention.

US Public and Private Markets

Apollo recently published a quick review of the dividing line between the public and private markets in the US. CNBC-style market narratives often discuss the difference between Wall Street and Main Street or how the stock market is not the same as the economy. It contains some quick data points worth sharing:

• Employment in S&P 500 companies is 18% of the total US employment. I’ve discussed in the past how elite technology companies can produce immense scale with relatively small teams. NVidia and Visa employ roughly 25,000 people each. This is not great for the inequality issue. Small and medium-sized businesses provide nearly all of the jobs.
• Less than half of all corporate debt outstanding is from S&P 500 companies.
• S&P 500 profits are half of the economy-wide corporate profits.
• Capex by S&P 500 companies, investments made in buildings and equipment, is just 15% of the US total.

When thinking about the state of the economy in your town or where you vacation, consider how divided that may be from the state of the S&P 500. Yes, everything is connected, but at this stage, the S&P 500 is roughly 25% weighted to big tech, and in the Nasdaq, it’s over 50%. Of course, this doesn’t address liquidity at all. Market values in the public markets also go up and down based on liquidity, whereas private markets don’t have a price until someone hires an investment banker to sell.

My other favorite divide in private vs public markets is the private credit market. There is too much demand for public credit and not enough supply (issuance). This has been true for a while now. As a result the yield on public credit is marginally better than treasuries in most cases. Due to 2008 financial crisis regulation, private credit moved out of the banks and into the asset market. It’s now a major multi-trillion dollar asset class that barely existed 10 years ago.