Welcome to the latest edition of The ROE Reporter where we delve into the dynamic intersection of technology and investing. In this edition, we turn our gaze towards the transformative power of artificial intelligence (AI) and its profound implications for the investment landscape. With each passing day, AI innovations are reshaping industries, revolutionizing processes, and redefining the very essence of how we interact with technology. From predictive analytics to autonomous vehicles, AI is driving unprecedented efficiency, accuracy, and insight across diverse sectors. Through real-world examples, we’ll explore how these advancements are not just shaping the future, but also presenting compelling opportunities for astute investors to navigate and capitalize upon. Join us as we unravel the complexities of AI-driven disruption and uncover the potential it holds for shaping the world of investing as we know it.
The above introduction might sound “off” in regards to our normal writing style and that is because it was written using ChatGPT with a one sentence prompt from us. We also fed the newsletter back into ChatGPT when it was done and asked for a title, which is where AI Ascend: Navigating the Investment Landscape was created. Playing around with AI tools like ChatGPT or Adobe Firefly will really open your eyes to the power of AI, which is the main focus of this edition of the newsletter. We’ll provide a primer on AI and focus on areas where we think real opportunities exist today, plus relevant updates for a few of our current investments.
The rebound in small-cap stocks started in late 2023 and continues to roll in 2024. We have spent a lot of time discussing the reasons for the opportunity and why this should be the start of the next 5-7 year cycle – low valuations, peaking inflation = peaking interest rates, and a historical amount of cash on the sidelines. (see our January newsletter here). We believe the next few years will be very strong for this segment of the market. Overlaying some of the innovations we’re seeing like AI, but also security and cybersecurity, healthcare tech, and grid tech, and future returns look very promising.
What is Artificial Intelligence (AI)
We’ll do our best to give a succinct overview of “AI”, but be aware there is a lot more nuance to this term we’ll get into today.
What we currently see in the world is mostly machine learning (ML) which is not quite the same thing as artificial intelligence. AI is being used as an overarching term with machine learning being under that AI umbrella. Machine learning can be thought of as advanced pattern recognition, or even just highpowered mathematics. A user feeds in data to train the “machine” (algorithm) which teaches it to perform a specific task by identifying patterns in data. In other words, using past and present data to decide on a course of action most likely to achieve a desired outcome. The more data, the more time in training, and the more compute power, leads to more accurate and useful outputs. Use case examples include predicting the weather, reading lips, analyzing x-rays or MRIs, etc. All of these can now be done faster and more accurately than humans. Examples of popular tools would be ChatGPT for generative AI, Siri & Alexa for virtual assistants, and IBM’s Watson for business data analysis & insights.
Currently, the world, and especially in the world of investing, Machine Learning is construed as AI, but technically it isn’t artificial intelligence. For the sake of simplicity, AI is used as a catch-all term. Actual artificial intelligence uses logic to reason, learn, and self-correct. Some refer to this as Strong AI where the machine is self aware and has its own emotions and beliefs1 . This is currently “Hollywood AI” and what we see in the movies. True AI is currently hypothetical but certain experts proclaim it’s near while others predict it will take decades or centuries. For purposes of discussion today we’ll refer to AI in its general context similar to the very broad terms of “industrial revolution” or “the internet”.
Early Impact
We’re still in the early days of AI but the main implication is increased productivity. Being able to do more with less. A few of the headlines above reference this, with some analysis saying industries on average will be 40% more productive in relatively short order.
This may be hard to fathom but there are already real-world examples with astronomical outcomes. University of Cambridge researchers developed an AI to find new treatments for Parkinson’s. It accelerates the initial screening process 10x, reduces costs 1000x, and already led to the discovery of 5 new compounds. Another would be implementing AI into robotics to perform routine and repetitive tasks – take a look at this Janitor robot here, or more impressively this 3 min video of the newly launched Astribot here. Just remember, this is currently the worst this technology will be and will only get more advanced from here.
We had an interesting internal discussion on how to compare what this might look like in the future using an historical example. What we landed on was what farming looked like in 1890 versus what it is today. As technology improved so did productivity. In the late 1800’s it took 83 labour hours and 2.5 acres of land to produce 100 bushels of wheat2 . With new technologies, it now takes ~2 hours and 0.6 acres of land to produce 100 bushels of wheat2 . A fraction of the time and a fraction of the inputs. What was even more impactful was the second order effects of this productivity. In the 1800’s over HALF of the ENTIRE labour force were farmers. Today farmers make up 1.2% of the labour force2 . Farming technology freed up time and energy, which led to new industries and further innovation.
Much like farming, for AI there are a multitude of second order implications. Performing computing at this level requires a lot of electricity. Where is the power going to come from? Is the power grid ready and able to transmit all this power? (it is not). Who and how will we be protected from those who will use this technology against us? (cybersecurity, physical security).
There will be innovation to help solve some of these issues like NVIDIA’s recent chip announcement that requires 25x less power3 or scientists in the UK who developed a new optical processing device that opened up new wavelength bands which set a new record at 301 terabits per second which is 25,000x faster than the world’s average internet speed and 2,000x faster than current 1Gb speeds being offered through fiber4,5. There will continue to be innovations like these but there will also have to be A LOT of investment into the infrastructure. Significantly more data centers, more chips, more energy, more hardware, more software, and more people to construct and install and monitor all of this.
Investments
We’ve gotten this far into the newsletter and one might now be expecting examples of AI companies we’re investing in. This is where we would like to temper expectations. Just like the internet bubble, there are A LOT of pretenders using AI as a catchphrase which ranges from deceptive to outright fraudulent. Just this month we’ve seen multiple “AI” pitches where we were left thinking “where’s the beef?” There are the obvious real choices like NVIDIA for the chips, Alphabet for search, and Adobe for generative content. That being said, these stocks aren’t cheap and their past performance does not guarantee their future. Where we have found true progress and real opportunity is in companies using AI as a tool to improve efficiencies, which is showing up in improved profit margins and accelerated growth (indirect monetization of AI versus direct monetization of AI charging explicitly for AI products/features). From an investment perspective, we would suggest thinking of AI as a tool to utilize versus an outright business model.
Tantalus (GRID)
- Tantalus is an interesting company that is potentially approaching a breakout moment. They have been around for a long time, supplying hardware to utilities and are consistently profitable. That is until a couple years ago when they invested in developing a new piece of technology. The power grid in North America is extremely outdated and nowhere near ready for the upcoming demands of AI. Their TRUSense Gateway is a piece of hardware and software that enables communication and data analysis from the substation, to the home meter, to the smart home appliciances. This behind the meter control system will allow utilities to be able to monitor power precisely, protect transformers, and take smart home appliciances to the next level.
- Just last week they received certification that now allows them sell to utilities. Eight of their existing utility customers partnered with them in helping develop this technology and it appears like they are now in the initial steps in rolling out this product. If the rollout goes well, just these eight customers can grow GRID’s revenue more than 3x.
Converge (CTS)
- Converge sells equipment to you like NVIDIA chips, offers services like IBM Watson, provides the necessary services to get you up and running, and maintains this infrastructure.
- They are executing on the massive amount of capital being spent on technology upgrades, AI being part of that. For these businesses that want to implement new technologies, they need to buy from a provider plus then employ a service provider to come in and install, implement, and bring to fruition. Converge offers both sides of this equation and as their service business expands so do their margins.
- The stock is still trading extremely cheap, with very strong organic growth and impressive free cashflow.
See full newsletter report here.
