Coffee Shop Thoughts: 11/27/24 - Tariffs, Mindsets, AI Knowledge Graphs, Transportation
Another brain dump in the same week?! Yep, I took today off from work and kids only to catch up on work. Classic.
Per usual, I’ve had new thoughts since I last wrote. So, let’s venture Beyond the Yellow Woods!
Tariffs and Trade Wars
This is a highly complex topic that I think so many people overgeneralize and don’t look at the semantics. Many people I see just make blanket statements like “this will be inflationary” without playing out multiple scenarios of how this policy could play out. You have to look at the complex interplay and totality of the policies.
I happen to be of the opinion that, given the current geopolitical environment, tariffs are going to be a good thing for USA for a variety of reasons. We did this before back in the presidency of McKinely in 1890. This raised import taxes between 38% and 49.5% on the majority of imported goods with some goods reaching even 70% import tax. Furthermore, we took it even further with a duty tax increase under the Dingley Tariff Law which increased the tax to 57%. A massive increase.
So, what happened?
It changed the economy to be an export-driven economy creating a significant demand for domestic industrialization. This helped spark a significant increase in domestic productivity and reduction in unemployment as well as decreasing the income inequality gap to the lowest levels in 15 years. It also helped generate a significant rebound from the economic depression of 1893-1897. Some cons were that consumers in the USA experienced higher prices during this transition but those were eventually and largely offset by an increase in wages. Another con was that this led to a deteriorated state of foreign relations but, again, this was ultimately offset by the coming World War I. Lastly, it increased the federal revenue by $2.1B (roughly $72.85B adjusted to today).
In a nutshell, it was a transition but one that didn’t break the economy and ultimately positioned the USA to be in a much healthier position going into the 1900s. Not only that, but it really positioned us well to be a dominant supplier for WWI. We played the long game and it worked well for the USA.
We obviously have a different economic environment today but there are parallels. An aspect I keep thinking about is that if we impose tariffs we will see a significant uptick in inflation due to the nature of our globalized supply chains. I disagree with the assumption that inflation is due purely to poor monetary policy and printing money. Yes, that certainly can be an origin point for inflation but inflation isn’t just a single-point phenomenon.
In today’s environment, if we increase tariffs on goods that we import, we will certainly see an increase in inflation. However, we can shift the supply chain behind that to other countries with which we have better economic relations. A key example of this that I believe will come true is the shift of off-shore manufactured goods from China to Argentina. Argentina’s current wages are roughly 25% cheaper than China's, and that’s before Argentina’s coming economic revolution and new/expanded trade deals with the USA.
Outside of that, one of the other policies being floated is removing income tax. This, to me, is a sort of round-robin way of never experiencing “inflation”. The way this works is that if we add significant tariffs, we would see the price of goods go up but the USA consumer would have substantially more purchasing power, offsetting the cost of those imported goods. For example, if I make $100k/yr, I’m paying ~$17,053/yr in income tax. So, 17% or $1,421/mo. That’s a lot of money.
Removing the income tax would also be a massive economic stimulus, pushing equities higher which helps retirement as well as spur domestic productivity with new businesses.
The obvious next question is what happens to the federal government’s revenue if this were to happen? Wouldn’t we be operating at an even larger deficit?
Not necessarily. If you take the rough math from the McKinley presidency, we’d see an uptick of roughly $350-450B in federal revenue. Not enough to offset our deficit but this is where another initiative comes into play: Department of Government Efficiency (DOGE). The federal government spends roughly $6.7T/yr and growing. A key factor in this is purely just bullshit spending. We all know it. We spend money on shit that we absolutely shouldn’t be. I believe there is an opportunity to cut this spending by 50%, maybe more.
Between significant federal expenditure reduction, a significant increase in import taxes, and a significant increase in domestic productivity in the form of new businesses that pay corporate taxes, there’s a path to getting close to balancing the budget. This is before we do anything really crazy with monetary policies, such as potentially pushing for a sort of Plaza Accord 2.0.
My point in writing all of this is to not throw the baby out with the bath water. On the surface, it may seem like a dangerous policy. However, I think that is woefully underselling the total economic plan that has a real potential to unleash our economy in spectacular ways.
Corporate Mindset
I’ve been in tech for over a decade now with a lot of high-growth companies. One of the consistent challenges leadership faces is maintaining growth and delivering incredible results.
I’m not saying that everyone should fall into the category of “hypergrowth” or whatever. But, here’s a thing to consider.
What happens to the mindset of your leadership team when you ask the following question:
“What would it take for us to 5x our revenue in the next 12-24 months? What would have to be true for it to happen?”
I’ve found that asking this line of questioning generates an intense amount of clarity about what is truly important in the business and what actually needs to fundamentally change about the business and industry. When you push it far enough, it’s a profound mindset shift that forces folks to suspend the current reality constraints that are often self-imposed and instead create a fictitious reality where the impossible is possible.
The funny thing about business is that this is how you win. You create fake worlds where you achieve an ideal end state that stems from your mind. And then? You go build that. You bend the arc of reality in your favor.
We need to be doing this so much more.
Knowledge Graphs for AI
I had a call with a lead investor of an AI startup earlier this week that confirmed a lot of my opinions about the desperate need to incorporate knowledge graphs into the broad LLM conversation.
The company we were talking about is building an AI-native competitor to Solidworks or Autodesk. Think AI-generated 3D modeling for industrial design (auto manufacturing, aerospace, defense, etc.). These are industries where precision is non-negotiable and proprietary knowledge is paramount in creating an asymmetric advantage against your competitors.
They are building a platform whereby you have a co-pilot collaborative AI that helps you produce 3D models with mathematical backing (eg. linear stress testing). Both the modeling and the testing are done relatively manually right now. Perhaps more importantly is that there is a lack of scaled experimental design - meaning, instead of testing 3 different versions, you test 100s simultaneously.
One of the key elements of their business is to back their platform with the proprietary data warehouses of these companies since these companies have a massive corpus of information relative to their product lines, business models, and industry nomenclature. The key takeaway here is that these data warehouses are not public and no public foundational model has the depth/breadth of training data to get close to what these companies have.
We talked about knowledge graphs as one of the critical elements missing and that this company aims to have a portion of their product aimed at creating a solution for it. Again, the reason you would want to do this to back a foundational model is the proprietary semantic reasoning found within knowledge graphs.
The other element that was confirmed for me is my belief that innovation from this next wave of AI will stem only from 2 places in the stack: The application/UI layer and the foundational modeling layer.
The application layer means venturing beyond just a basic chat-style interface and moving more towards an intelligent co-pilot or collaborator with visualization building in real time. I think some great examples of this are Blender, Unreal Engine, and Solidworks.
The foundational modeling layer isn’t simply just creating new foundational models but creating a vertically integrated solution by the industry for backing these foundational models (eg. Llama3) with the ability to inject reasoning into them. It’s creating more control over how the AI thinks, pointing its perspective at a problem in a specific way based on certain injected contexts.
Both are required to generate leaps in innovation. But, once we get there, building the physical future will be akin to dreaming. Anything you can think of can be built.
The Next Evolution of Transportation
One of my close friends is becoming a private jet pilot and we’ve been talking a lot lately about the market. One of the interesting things that I’m seeing is that I believe there are very soon going to be 2 new widely available mediums of transportation.
VTOLs and Rockets.
We are so close to a more efficient and safer form of mid-range transportation and that’s with VTOLs. Most notably, the one leading the charge here is Archer Aviation. When you think about the scales of transportation, we have bikes (slow, short-range), cars (somewhat fast, mid-range), trains (slow, long-range), and airplanes (fast, long-range).
With cars, the most significant downside is that they are point-to-point transportation solutions but require rigid predefined paths of transportation which increase the transit time significantly. VTOLs present an opportunity to do near-point-to-point transportation for short to medium ranges. Archer Aviation is pushing for a taxi service and will start to open up routes in San Francisco, Los Angeles, and Chicago over the next few years. These routes will present an extremely fast and efficient way to get “close” to your destination from a “near” pick-up destination. I suspect that we will see this increase significantly over the next decade in popularity and proliferation, particularly as the FAA gets overhauled (the biggest limiting factor is regulation from FAA). It will be much cheaper than a plane but more expensive than a car, serving those who want to pay a premium to get their time back - which is a lot of us.
At the other end of the spectrum is rockets which will become the next evolution of private jets. Why the fuck would I want to charter a G600 at $10,000/hr that takes 14.5 hours to get from LA to Dubai at a cost of $145k when I could hop on a rocket for $50k and get there in 45 minutes?
For the elites of the world, this is where we are heading. It’s a no-brainer. The byproduct of this is that as demand increases, we will see costs come down making it more widely available. Not only that but we may even see needs for things like a “space terminal” where rocket capsules carrying passengers dock at the terminal for a short period of time and transfer to a re-entry capsule for landing. This would open up an entirely new and robust economy for humans.
That… and it would be cool as hell for us to finally achieve that future sci-fi state we were all promised.
I actually think this is going to start happening within this decade. Again, as FAA regulation gets retooled, we should see a significant change in the economic dynamics of the rocket launch industry to open up new commercial markets for us.
Don’t sleep on transportation. There’s a revolution coming.
Alright, that’s it for now. Drop a comment if you have any thoughts. I hope everyone has a Happy Thanksgiving!
- Ryan