Noahs' Arc 2023 Q2 Newsletter
AI, of course... but maybe not how you'd think...
Welcome to the Noahs’ Arc Q2 2023 newsletter. We’re talking about how AI will revolutionize not just the work you do on excel, but also the cost to produce a barrel of oil, how garbage is collected, and how utilities in flood zones respond to natural disasters. We’re excited about the tangible impacts of change, not just the fluff and hype. This is one of our longer letters by design. Buckle up.
No aspect of this material is intended to provide, or should be construed as providing, any investment, tax or other financial related advice of any kind. You should not consider any content herein or any subsequent services provided to be a substitute for professional financial advice. If you choose to engage in transactions based on the content herein, then such decision and transactions and any consequences flowing therefrom are your sole responsibility. Noahs’ Arc does not provide tailored investment advice to any person directly, indirectly, implicitly, or in any manner whatsoever.
Executive Summary
AI in Boring Businesses - AI will fundamentally change how 20th century businesses operate, making some unexpected lines of business more stable, profitable, and moated in the long run.
Investing in AI - A lot of investor dollars are being thrown at big tech; some legislation that just passed in the House may affect the underlying market dynamics encouraging this.
Personal Note - Noah and I completed our senior years! We’re staying at a hacker house in Austin for the summer to build the fund and to work on our AI division, Ultima Insights.
AI in Boring Businesses
Artificial intelligence is to 2023 what Web 3 was to 2020. All the talk of sustained heightened interest rates, a flight to quality, and the hunt for stable cash flows don’t seem to matter for most AI stocks and startups.
While some valuations are likely inflated, there is long term promise in AI. This is a revolution that we believe will bring more value to the customers of these AI companies than to the shareholders of the firms.
To be clear, we think some AI companies will make a lot of money. However, this is quickly becoming a crowded trade. We think some of the best places to obtain alpha from the AI revolution will be found in “boring businesses.”
Take British Petroleum (BP). It's a classic oil company riddled with all of the complexities of a modern supply chain. It operates in a competitive, commoditized market – while management can’t easily control sale price, they can optimize for a lower cost basis. Leveraging AI, that is exactly what BP did.
BP noted in their last quarterly conference call that they were able to use Palatir’s AI system, dubbed Foundry, to help cut the costs of barrel lifting in their Oil operation from $14/barrel to $6/barrel. That is more than a 50% cost saving in a traditional, boring, 20th century business – talk about margin expansion!
In our view, the way many “boring” businesses will win in the future is by leveraging their knowledge bases & data to drive exceptional operations. Right now, a company’s “knowledge base” and best practices reside mainly in the brains of each of their employees. The challenge in making a sustainable business can be viewed as the challenge of making a sustainable system. This task has, historically, fallen on humans.
No longer.
Now, the best processes and enhancements will be largely devised by AI and remembered by AI. These knowledge bases will become indispensable.
Charlie Munger only enters trades in which he knows more about the underlying situation than the counter party. For many commoditized companies, AI can take the surplus of knowledge and data laying around to hone their information edge.
On their last conference call, Waste Management (WM) discussed their use of AI to optimize delivery routes by a driver’s skill level. This is the sort of care and attention that a small mom and pop garbage company may have been able to figure out for their 5 or 10 drivers; however, this attention to detail is something often lost when a company scales as Waste Management has. AI is flipping the narrative; now, WM can lean into its data & scale, using it as an advantage of a different sort than before. Properly processing their information creates systems that are hard for other players to compete with.
This exasperates the winner takes all effect that accrues to the largest, most competent player. With the knowledge being contained in AI systems vs. human brains, it will be harder for market entrants to compete. More stable profits may result in the market rewarding these firms with higher PE multiples.
On the supply side, profits for generative AI models may decrease in the long run (niche models and firms could still likely have their place, though). Take OpenAI, for example, who just this week cut the cost of their GPT3.5 AI model by 25%. LLMs (large language models) may themselves become a commodity. Moreover, AI will enable faster generation of software, including competing LLMs, meaning profits in many parts of the software industry could go down due to lower costs to replicate.
As we mentioned in our Q3 2022 newsletter, we believe that there will be real increased interest in hard asset businesses as we move focus from the bit back to the atom. Efficiency improvements via AI are a strong mid term impetus for this trend as companies with real assets increase their moats.
Finally, check out this video about Palantir helping an engineering company, Jacobs Solutions, use AI to enable a utility company to control flood zones. It's fascinating, and the cost savings are tangible.
Investing in AI
Other than the efficiency gains from AI, we’ve been watching what proportion of created value the AI companies capture. We think there is a risk of much of the business or LLMs being commoditized.
Interestingly, if you look at the S&P 500 YTD, taking out 7 stocks ($NVDA, $AAPL, $TSLA, $MSFT, $GOOGL, $META, and $AMZN), as of 12pm est on 6/23, the S&P 493 is now flat YTD. The S&P 500 is up 14% YTD. The market’s gains this year are being largely driven by tech, and we seriously wonder what portion of those gains are caused by the AI craze. Assuming that even a third of the price increase is attributable to AI, AI must have the power to multiply the profits for these firms, right?
We think it’ll be a boost to profitability. But again, we think it’ll be the niche applications, not the general ones, that’ll produce the real profitability boost.
Google infamously pointed this out last month when the “We Have No Moat” memo leaked. Essentially, the growing parity between generative AI models like GPT, Anthropic, PaLM (Google’s model) and LLaMA (Meta’s model) risk eroding the profits that the models seem to be promising for their creators. If multiple AI models can answer the questions you have, and can all answer them accurately, then what is the difference of using one model over another?
Most people don’t pay for ChatGPT; the average individual is not used to paying for information search and retrieval. It's largely enterprise firms that pay for access to AI models like GPT. And, like we mentioned before, OpenAI is cutting the cost on some of these premium models aggressively.
Even if these tools do replace traditional internet search, unless ads can be effectively placed inside ChatGPT & competitors, these tools may become loss leaders for business models.
Furthermore, investing in AI may be about to change. Currently, in the US, if you are not an accredited investor, the best way to play the AI trend is to buy expensive AI stocks such as NVDA, which is currently trading at 219 times earnings.
In Congress, legislation has passed the House allowing anyone to become an accredited investor – you’ll just have to pass a test. Currently, reg CF offerings (the only existing way to take money from investors who are not accredited) are the kiss of death for startups; if this regulation passes, the supply of money available for more kosher securities offerings will increase for startups. Angel investing is trendy, and spreading risk over multiple angel investments may offer a better risk/reward ratio than buying high flying AI tech stocks. Bloomberg even just reported the other day that many startups are starting to look like a bargain in the private markets.
After all, who wants to pay 219x earnings for a trillion dollar chip stock trading at 40x sales when you can angel invest, spreading your risk and maintaining the potential to 10x your money?
Personal Note
Noah and I just completed our senior year!! It's crazy how fast the 4 years at the University of Michigan flew by. We’re in Austin for a couple months to live in a hacker house organized by our friend Bobby as we focus on our fund and build the AI side of our business, Ultima Insights. We’ll return to Ann Arbor in the fall.
A little different than the house in Silicon Valley, but with similar energy. It’ll have a better ending than the TV show.
From the bottom of our hearts, we both genuinely appreciate everyone who has stuck along reading this newsletter as we’ve been pursuing these ventures. Two years ago, all we had was a newsletter. Now, we have a very real opportunity at our hands that allows us to pursue our ventures full time.
The amount we’ve learned so far has been jaw dropping for us, and we look forward to continuing to grow & build.
Till next time,
Noah & Noah
No aspect of this material is intended to provide, or should be construed as providing, any investment, tax or other financial related advice of any kind. You should not consider any content herein or any subsequent services provided to be a substitute for professional financial advice. If you choose to engage in transactions based on the content herein, then such decision and transactions and any consequences flowing therefrom are your sole responsibility. Noahs’ Arc does not provide tailored investment advice to any person directly, indirectly, implicitly, or in any manner whatsoever.
Very proud of you guys.