I think about this scene from Parks and Rec pretty much all the time:
It has nothing to do with Applications or Zero Interest Rate Policies (ZIRPs). I just want you to think of me and Aziz every time you go out to dinner, for the rest of your life. If you’ve been to dinner with me, you know I always get apps and zerts.
TLDR/ExecSumm:
ZIRPs and money printing created an environment where there is a ton of inflation. In order to curb inflation, growth must be curbed. This creates a dangerous cycle where no progress is made because growth is expensive. AI applications can hack that cycle and create value and grow the economy without relying on cheap money.
ZIRPs
So what does have to do with Apps and ZIRPs? Pretty much everything else going on in the world over the past decade. Here’s a quick refresher on the Fed rate over the past 70 years or so if you have been sleeping under a Dwyane Johnson:
As I talked about in my 2023 predictions article, I think we are going north of 5% and then waiting to see if things break (meaning: job loss, demand shift, decline in prices/inflation). Depending on if things break, how quickly and badly they do, will determine how long we stay north of 5%, how far rates comes back down, and how long they stay at their new normal (which most are considering to be 2 or 3%).
Since nothing is ever binary, and the US economy is a panacea of interconnected activity, we can’t say for sure what the true cause and effect of any economic boom or bust is, because their are millions of factors for each cycle. Sure, economists will confidently tell you exactly what happened and why, but as the old saying goes, “economists have predicted 9 of the last 5 recessions”.
Anyways, even a real life Dwayne Johnson could tell you that part of the current recession can be attributed to ZIRPs. Cheap money being blasted into all sort of things that (in hindsight) didn’t deserve to be funded brought forth a particularly frenzied investment environment. As I have pointed out in many previous articles, part of the reason Crypto speculation got out of control is because excess money spilled over from Stocks to PE to VC and eventually into the blockchain. Low interest rates on borrowing means that investors can be riskier in their portfolio construction. COVID and the government stimmies also exacerbated the trend, yadda yadda, and much of the market cap that was propped up in 2020-21 has come back to Earth, we know.
The ZIRP environment from the past 10 years has been so successful, that the Fed still can’t seem to curb inflation. Generative AI might be here to bail us out, though.
As a gold rush of new AI companies are being incorporated as we speak, we can presume that the value they unlock is going to have some part to play in how the next phase of the economy unfolds. After all, if the goal of the FED is ultimately raising rates to curb inflation, and the only way to curb inflation is to halt demand, and the only way to halt demand is for jobs to decline, well.. AI is the perfect tool to bring about this change in a non-ZIRP environment.
One other salient point on ZIRPs. If you zoom in to just the last 10 years, you can see where COVID caused the immediate plunge of 2.5% down to 0.
This dip and subsequent recovery looks just like the blip that COVID caused to the S&P 500 in early 2020.
My point being, in both of these instance, we were on a path upwards, and the exigent circumstances were just a setback. The S&P was on its way to 5000, and the Fed rate was on its way to 5%. In both instances, we landed where we were supposed to, but the path was not direct. My general investment thesis is pretty bland. Like an expanding universe, as GDP goes up, so does the market. Increase your time horizon and everything looks amazing.
Considering that Tech is largely deflationary, it says a lot that we had such inflation over the past few years. It means not only did we print too much money, but that money was not put to good use in creating technology. What revolutionary products have been added in the past decade? Did Facebook and Google and Apple capture our attention so profoundly that we forgot to innovate? They themselves were printing so much money that they didn’t even need workers anymore:
Of course that did not stop them from massive hiring frenzies, but it’s their money so they can do what they want with it.
What happens if the Fed rate continues to climb and nothing breaks? If we are where we are supposed to be, then 5% ain’t nothing to be scared of. No jobs are being lost. Margins are too good everywhere. Plus, people aren’t getting out of bed to work unless it’s for 6 figures.. so a lot of people aren’t getting out of bed to work. But that’s all good when you’re a stay at home girlfriend or your parents are paying for your life. Nothing breaks, the Fed rate goes up.
Company valuations are affected though. If the cost of borrowing is up, investors need a safer return, which portends additional scrutiny. This is bad for VC and Tech. They need that cheap money so they can place risky bets on cool shit like Crypto and food delivery services. And every once in a while a ho-hum thing like Google or Facebook or Dropbox or Spotify or Zoom or GitHub or Snowflake or WhatsApp will emerge out of the chaos.
The conspiracy theorist in me wants to think that ChatGPT was strategically released just as Crypto was phasing out so that all the hundreds of billions of VC dry powder sitting on the sidelines could be put to use, stimulating the economy in the only way that tech can: by being massively deflationary.
Unlike crypto, the hype around AI isn’t all rampant speculation. Crypto did and does have a massive value potential, but the value created for VC in that space was always going to be getting on on the ground floor of the token creation. As network effects coalesce around rebuilding the entire web infrastructure, the value capture is in owning a bunch of Disneyland bucks before Mickey Mouse is created.
Money is stingier now, so the value unlock has to be more real. While DeFi threatens to take jobs away from banks with AMM bots, it requires that the entire financial system be uprooted. AI is more incremental, more tangential. It threatens to take away jobs from accountants and paralegals. It can happen one by one, it doesn’t need institutional collapse. But how does that incremental change take place?
Apps
AI and ML technology has obviously been developing for a while now. GPT3 has been writing paragraphs for years, it is just the chat interface that was added that made it consumable by the public. And therein lies the beauty of the Apps.
I discussed leveling up in The End of Knowledge, and going from 0-1 in code in media in The AI Article. The thing that makes all of this possible is the interface between the AI and the user. As I wrote in Middleman America:
“Advances in AI will undoubtedly create another unlock in the near future and the entire world will bubble up yet another layer removed from where we are now. There will be untold levels of value creation, with more valuable Middlemen.”
As it stands right now, OpenAI is the hub of the entire ecosystem. I wrote last week about how the corpus of content is king, but the content is nothing if you don’t have a functioning AI model on top of it. The good news is that OpenAI has opened the kimono and is allowing developers access via an API key. This is actually nothing new, as apps like Jasper have been utilizing the GPT3 dataset since 2021.
The AI and ML models will undoubtedly grow bigger, and better, and utilize both more broad and bespoke datasets, but that’s the hard infrastructure work that can’t and won’t be democratized in the short term. The barrier to entry is too great. You need 100 engineers working around the clock, with a blank check to pay for compute. The gold rush right now is getting customers in the door, and the way to do that is Apps.
The path forward for entrepreneurs is to build that next killer interface that makes interacting with the AI easy and impactful. The stage is already set where knowledge work is primed to be outsourced. People are expecting it. All businesses are potentially able to make more margin due to less headcount.
Sure, ChatGPT can analyze a spreadsheet if you feed it the information properly and prompt it in precisely the correct way. It gets you the answer you desire, but for complex problems or detailed artwork it can be extremely laborious. If it takes you even 25% less time to perform a task, it is not worth it. A product needs to save you 99.9% of your time for people to adopt it.
This all comes at the interface level. Apple’s greatest innovation wasn’t the computer, but popularizing the GUI. Right now, interfacing with OpenAI via ChatGPT is like interfacing with your computer through DOS. Prompting is like code, and needs to be abstracted away. This is why Prompt Engineers are a thing like Software Engineers are a thing.
Here is where I think the Chris Dixon mental model on Crypto is applicable, which was that there are two major design phases in technology: skeuomorphic and native. Skeuomorphic means that the new technology looks and feels the same as the legacy item it is replacing. Think: the calculator app on your phone looking like a calculator. This is a good stepping stone because it is hard for designers to visualize something that doesn’t exist yet. They are creating something of utility and have only the incumbent design as an example. The native era is when the underlying tech gets completely reimagined and the skeuomorphism is abstracted away, never to be thought of again.
Just like with Crypto, there are going to be a lot of AI ideas to replace legacy tasks in the exact same way. Just putting something on a blockchain constituted an idea. With generative AI, it’s much the same thing. Think of a job that can be automated away, and then use AI to make that job easier. This will provide a lot of value I am sure, but still will not evolve into the ultimate prize which is creating something native to AI that was not previously imagined and only AI can make possible.
The current gold rush of generative AI companies all linked up to the same database is the natural first step. The underlying AI models, hardware, and compute will ultimately control the stack much like Apple does AppStore and AWS does with servers. That doesn’t mean that there aren’t billion or trillion dollar companies that can be built off their backs. Facebook was wildly successful despite being held hostage by the need to exist on a piece of hardware controlled by another company. Also, Apps are the tip of the spear in this scenario. The rest of the stack might ultimately capture most of the value, but that value won’t be captured without the Apps. The iPhone wouldn’t be as wildly successful without Facebook, TikTok, YouTube, etc.
Note: this visual is only the value capture from the server-side, so to speak. The client side is a little bit more of a curveball. Executives will get to pick and choose where and how much of the margin savings goes in their organizations.
This upcoming vintage of AI companies has the potential to do a lot of things. Namely, create a lot of value in businesses that couldn’t otherwise be optimized away. Technology is ultimately deflationary. As thousands of entrepreneurs line up to create that interface, there will be a few hundred that are sticky, which will affect hundreds of millions (or billions) of users on the client side.
So this means massive Main Street layoffs, an end to inflation, and a Fed movement closer to ZIRP territory, right? Could be. All of this sounds great to investors and entrepreneurs, but what about those people whose jobs are going to be so callously automated away?
While others may be doom and gloom about the impact of AI and its affect on humanity (a dystopian Brave New World, leading us into UBI and Soma addiction), I am more bullish on the ingenuity of humans to rise to the occasion.
Consider this: intelligence is a proxy for wealth, or at least a version of wealth. It is not a perfect correlation, but broadly speaking the more intelligent you are the more money you will make. There are extreme outliers in the case of poverty (the zip code you were born in) and nepotism/crony capitalism, but by and large a certain degree of intelligence allows you to navigate the world more advantageously. Let’s assume no correlation, however, and just use intelligence interchangeably as a metaphor for wealth.
We are at the most extreme version of both wealth and intelligence inequality at any time in our existence, and it is due entirely to technology.
Some people are really really smart, they can do great things when they apply themselves and make massive impact. Other really smart people crash and burn of course, the same way some people inherit tons of money (or win the lottery) and blow it on helicopters and cocaine. But while the MIN human intelligence (or wealth) rises linearly, MAX human intelligence rises exponentially over time. It is the top 15% of society that raises the bar and pushes things forward, not the straggling troglodytes.
This progress is quite observable in daily life and recent history. Outlawing murder, abolishing slavery, etc. Humans weren’t always so civil, and human history is not that long. You wouldn’t think that from the vitriol you see online and in the news, but that is not reality. Most people are extremely kind and generous. The trajectory of humanity is quite vertical when you zoom out. I’m long term bullish on humanity the same way I am on the S&P 500. Displaced workers will be displaced at the absolute best time in humanity to be displaced. Human ingenuity compiled with the increased leverage available through technology can create so many more interesting things in the world. The mind can shift from trying to act like a computer to acting like a human, and it will be exciting to see what sort of shift that entails for society.
I am reminded of a famous quote from one of the top minds in history (talking about alcohol but can be applied to AI):