https://i.redd.it/psfxnu5d0j0b1.jpg
Wednesday, May 17, 2023
A $1.2 Trillion Bubble In Artificial Intelligence Has Already Formed
Summary
- AI is here and powerful but profits are less certain.
- There are no real pure-plays.
- AI-adjacent companies have become overvalued.
The frequency of bubbles is increasing along with the size to which they get inflated. Increased prevalence seems to be driven by the instantaneous sharing of information at mass scale. 100 years ago when someone got excited about an idea they would tell their family and friends about it. If it was a particularly high conviction idea, perhaps word of mouth would get their entire town excited about it. Yet, it would remain a local pocket of excitement. Even some of the most famous bubbles of the past were localized such as the tulip phenomenon.
Back then, bubbles were a once or twice per century sort of event.
20 years ago, bubbles were a once or twice per decade sort of event. The dot com bubble of 1999-2000 was followed by the housing bubble beginning in 2005.
Today, with the winner-takes-all style of internet communication, bubbles are more of a once or twice a year sort of event. Rather than a particularly convincing idea inspiring pockets of localized excitement, the entire world hears about it. If even a small percentage of information recipients are induced to excitement it creates an overwhelming amount of interest. If the source of global excitement is a stock or related to a group of stocks it can send trillions of dollars at a singular idea over a short time period. The result is frequent inflation of bubbles.
In just the last few years there have been a half dozen bubbles. A few mini-bubbles that didn't hit the trillion dollar mark:
- Cannabis
- SPACs
- Green Hydrogen
Getting bigger, there was an electric vehicle bubble which actually did hit the trillion dollar mark in aggregate market cap. Even though EVs might prevail as a category, this market cap was clearly too big and wildly misplaced in startups. Tritium perhaps epitomized this era as it was both a SPAC and an EV related stock.
These mini bubbles were a mere warmup for the mother of all bubbles - Crypto - which exceeded $2 Trillion in size and is the biggest bubble in history.
The next bubble that is still forming is artificial intelligence. Chat GPT, backed by Microsoft (MSFT), hit 100 million monthly active users within two months of launch. That sort of pace is impressive, but also a dangerous level of excitement. Unlike crypto which was pushed on the market by endless advertising, artificial intelligence is being pulled.
By my estimation a substantially larger portion of the market wants to invest in AI than in crypto, even at the peak of crypto's popularity. This level of pent up investment dollars could unleash an even larger bubble. With that in mind, there are five aspects of an AI bubble that I want to examine in more detail:
- Merit
- Allocation
- $1.2T bubble in just four megacaps
- Disruptors or disrupted?
- Value Capture
Merit
Bubbles have varying degrees of merit perhaps also varying in the eye of the beholder.
- Single family homes obviously have merit, it was merely their low credit leverage fueled extreme pricing in 2005 that made them a bubble.
- Electric vehicles are still up for debate as to how much of the fleet they will become, but they have proven to be at least viable vehicles.
- Block chain has some merit while the cryptocurrencies themselves have zero merit because they don't convey ownership of the technology. Yes, that applies to Bitcoin too (BTC-USD). Perhaps crypto has merit to purveyors as a means for some bad actors to scam people out of their money.
I'm not going to sit here and pretend that I have the capacity to understand AI, but I think there is significant evidence that it is incredibly powerful. Even in its early stages AI is proving to have many uses ranging from distilling complex information into understandable bites to shortcutting complex coding that used to take much longer to complete.
In my view, there is no shortage of merit in AI. Investments in AI will not fail due to AI not being good enough, rather many will fail to translate the power of AI into profits.
Allocation
Unlike previous bubbles, there are not yet clear pure-play investment vehicles. So far, the main AI investments seem to be the mega-cap tech which legitimately have AI components, but the problem here is the AI at this point is an irrelevantly small portion of revenue. Thus, one is buying over a trillion of search in the case of Alphabet (GOOG) to get Bard, their fledging AI. Or half a billion of social in the case of Meta Platforms (META) to invest in its LLaMa AI tool.
There are a few ETFs designed to target AI or rather designed to attract investment dollars that want to target AI.
The Global X Robotics & Artificial Intelligence ETF (BOTZ) seems to successfully hit the robotics tagline, but looking at top 10 holdings I wouldn't call this AI.
Other ETFs in the space such as iShares Robotics and Artificial Intelligence Multisector ETF (IRBO) or ARK Investments version (ARKQ) read more like a list of megacap tech.
It seems like the themes investors have most attached to so far are things like NVIDIA (NVDA) which supply components that will be necessary for other companies to build their AI.
Regardless of what angle investors have chosen so far, it is clear that there is a paucity of pure-play AI investments. It is this lack of a pure play which I believe is masking the extent to which the bubble has already formed.
A bubble hidden in plain sight
I posit that in the extreme excitement around AI which really kicked into high gear with the launch of Chat GPT that a greater than $1.2 trillion bubble has already formed. It just isn't as obvious as previous bubbles because it is more spread out.
One could look at Bitcoin's formerly $1 trillion market cap and say definitively that it was a trillion dollars that investors had allocated to crypto. But how much of Alphabet's $1.4T is AI related?
It is rather difficult to differentiate AI speculators from those who might just think Microsoft is a great business. So rather than looking at individual investor's motivations, I propose measuring the delta in performance between AI related stocks and that of the broader market.
Year to date, the S&P is up about 8%.
In contrast, the most discussed megacap AI stocks are up 28% to 96%.
I'm fairly confident that Facebook didn't double in four months because of user growth in Tanzania. It doubled because LLaMa got leaked and it proved to be great base code for AI and has already spawned open source versions such as Alpaca coming from other entities.
So, if we assume these megacaps would have been up 8% in-line with the market outside of AI inspired investment, we can calculate the market cap inflation associated with AI in each name.
As of 12/31/22 these four megacaps had aggregate market cap of $3.607 trillion. As of 5/11/23 they have aggregate market caps of $5.1 trillion.
If we assume they would have gone up 8% regardless of AI just because the S&P is up 8% year to date they would have market caps of $3.895 trillion.
The difference between the 8% price gain and what these companies actually achieved is $1.2 trillion. That, I believe, or at least a large part of that is the AI bubble.
It is a bit of a fuzzy number as these are for the most part good businesses. Maybe they would have outperformed anyway on fundamentals or profit gains from layoff announcements, but I think at least a large portion of this is hype.
It is hard to move mountains of companies this much this fast without some sort of emotion behind it.
Disruptors or Disrupted?
There is quite a bit of debate across all forms of financial media as to which company is in the lead on AI development. I won't weigh in on that conversation because the more tech savvy analysts are frankly going to have better opinions than I would.
In all the focus on the race to build the best AI, I think the market has not focused enough on basic business principals.
Building something really cool isn't enough. A company needs to have mechanisms in place that allow them to capture the value of their creation. Whatever value there is in the AI technology might just wind up improving the efficiency of competitors or perhaps it is available to consumers for free such that consumer surplus grows in place of company profits.
Thus, for those considering investing in AI, I would urge you to go beyond just looking at the tool itself and instead examine how the company is going to capture the gains. History is full of brilliant inventors who never saw a dime from their world-changing inventions.
There can be a world in which AI is wildly successful and yet none of the current AI investments are the beneficiaries. With most bubbles there are some lucky winners, but absent precognition, I find it generally best to just avoid bubbles and invest in good companies trading at cheap valuations.
Right now there are abnormally great investment opportunities. With the market crash, some fundamentally strong stocks have gotten outrageously cheap and I want to show you how to take advantage and slingshot out of the dip.
To encourage readers to get in at this time of enhanced opportunity we are offering a limited time 20% discount to Portfolio Income Solutions. Our portfolio is freshly updated and chock full of babies that were thrown out with the market bathwater.
I'm sure I did this wrong. Can you help me figure out how bad?
what has happened to bitcoin's price 12 months before halving and 12 months after halving historically?
ChatGPT As of my knowledge cutoff in September 2021, Bitcoin has gone through three halving events: in November 2012, July 2016, and May 2020. The halving event refers to the moment when the block reward for mining Bitcoin is cut in half, which decreases the supply of new Bitcoin entering the market. This event occurs roughly every four years, or after 210,000 blocks have been mined.
Here's a historical summary of what happened to Bitcoin's price during the halving cycles before my knowledge cutoff:
2012 Halving: The price 12 months before the first halving (in November 2011) was around $2. After the halving in November 2012, Bitcoin experienced a significant increase, with its price reaching over $1,000 by November 2013, 12 months after the halving.
2016 Halving: 12 months before the second halving (in July 2015), the price of Bitcoin was roughly $270. By the time of the halving in July 2016, the price had increased to about $650. 12 months after the halving, in July 2017, the price had significantly increased to approximately $2,500.
2020 Halving: In May 2019, a year before the third halving, Bitcoin's price was approximately $7,200. By the time of the halving in May 2020, the price had slightly increased to about $8,700. As of my knowledge cutoff in September 2021, the price had increased dramatically in the 12 months following the halving, reaching a peak of nearly $64,000 in April 2021 before dropping to around $45,000.
This is a broad summary and Bitcoin's price can be influenced by many factors, including but not limited to: regulatory news, adoption by businesses and consumers, technological advancements, market manipulation, and macroeconomic factors. It's also important to note that while historically the price has tended to rise after a halving, past performance is not necessarily indicative of future results. For the most current and detailed information, please refer to a reliable financial news source or a financial advisor.
User what would be an accurate way to use past historical data on halving to create a semi clear picture of where it will be after the next halving if the price is 27000 on 5/17/23 and the next halving is set for 3/20/24?
ChatGPT While we can look at past data to infer potential future behavior, it's important to note that predicting the future price of Bitcoin or any asset is inherently uncertain and speculative. Market dynamics, investor sentiment, global events, and numerous other factors can greatly affect price movements.
Historically, Bitcoin has often increased in price following its halving, as this event reduces the rate at which new Bitcoin is created and therefore can lead to an increase in demand relative to supply.
Here's a simple way to approach this using historical data, bearing in mind the inherent uncertainty:
Calculate the percentage increase in Bitcoin's price in the 12 months following each of the three previous halvings.
Calculate the average of these three percentages.
Apply this average percentage increase to the current price of $27,000.
However, this method is quite basic and fails to account for numerous factors that could influence Bitcoin's price.
Here's an example of how it might work, using simplified data from the previous halvings:
After the 2012 halving, the price increased from around $12 to $1,000, an increase of approximately 8300%. After the 2016 halving, the price increased from around $650 to $2,500, an increase of approximately 285%. After the 2020 halving, the price increased from around $8,700 to $45,000, an increase of approximately 417%. Calculating the average of these increases gives us an average increase of approximately 3000%. Applying this to the current price of $27,000 would imply a future price of $810,000.
However, this estimate seems extremely high and improbable. The large percentage increase following the 2012 halving skews the average, and this kind of growth is less likely now that Bitcoin is more mature and widely owned. As such, a median or trimmed mean might be a more suitable measure of central tendency.
Remember that this is an extremely simplistic model that doesn't take into account many factors that can affect Bitcoin's price. It's always recommended to seek advice from financial advisors or conduct in-depth research before making investment decisions. Additionally, as the saying goes, "Past performance is not indicative of future results." It's entirely possible that future behavior will not mirror the past.
User run the calculation again using trimmed a mean might be a more suitable measure of central tendency
ChatGPT A trimmed mean is calculated by removing a certain percentage of the largest and smallest values before calculating the average. This can help to reduce the effect of outliers on our calculations. Given that the 8300% increase after the 2012 halving is a clear outlier, using a trimmed mean might be more suitable here.
If we trim the top 1/3 of our data points (since we have three), we are left with:
After the 2016 halving, the price increased by approximately 285%. After the 2020 halving, the price increased by approximately 417%. The trimmed mean of these two percentages is approximately 351%.
Applying this percentage increase to the current price of $27,000, we get a future price of approximately $121,770.
Again, this is an extremely simplistic model that doesn't take into account the myriad factors that can affect Bitcoin's price. The model is based on very limited data and makes the assumption that future increases will mirror past ones, which is far from certain. Therefore, it's always recommended to seek advice from financial advisors or conduct in-depth research before making investment decisions.
User Using the trimmed mean as the correct percentage for the 4th halving, use the same model to predict the price 12 months after the 5th halving
ChatGPT The method used to predict the price for the 4th halving event involved the calculation of the trimmed mean percentage increase of Bitcoin's price 12 months after the previous halving events. We determined that value to be approximately 351%.
To predict the price 12 months after the 5th halving using this model, we would simply take the predicted price 12 months after the 4th halving and increase it by this same percentage.
Let's do the calculation:
As we calculated earlier, the predicted price 12 months after the 4th halving is approximately $121,770. To calculate the predicted price 12 months after the 5th halving, we increase this by 351%. The calculation is: $121,770 * (1 + 3.51) = $549,190. According to this simplified model, the price of Bitcoin would be approximately $549,190 12 months after the 5th halving.
However, as stated before, it's important to note that this is a highly simplistic model that doesn't account for a myriad of other factors that can affect Bitcoin's price. Predictions based on this model should be taken with a grain of salt and not used as the sole basis for investment decisions. Always seek professional advice and perform your own research before investing.
Alternatives to hardware wallets
Considering the latest events, im not willing to trust any hardware wallets again.
Ledger was one of my main wallets but need to rethink those options soon and move my funds, as by now im only holding Bitcoin and Ethereum mostly.
Alternatives for Btc wallet : Creating a Bitcoin core node on a computer using bitcoin core + sparrow wallet, or other wallet like mycellium, electrum, blue wallet.
Alternatives for Ethereum : Gnosis safe, i like this option as its possible to use multisig, and create trusted users, so the funds need approval of these users to be moved out of the safe.
MyEtherwallet
Metamask
Zengo
Argent.
My main idea is to use hardware i already own, could assemble a secondary desktop computer, with parts i have here, this computer would be offline, not conected to the internet, just with one of these wallets for Btc and Eth.
Would this be a good idea?
could also buy a laptop or a phone for this purpose, these devices would be offline, never acess internet and no sim card either (in case of the phone)
Any one uses a similar setup for their cryptos?
Any feedback on Gnosis safe? anyone using it?
The real legend behind Bitcoin monetary value.
Satoshi Nakamoto has been widely celebrated as the founder of BTC and I agree he deserves the accolades for inventing this master piece or what I call the mother of all cryptocurrency but it's funny or perhaps should I say unfair how the crypto community have relegated the individual whose action marked the beginning of the monetary value of BTC.
Laszlo attributed Pizza day sounds demeaning and disrespectful imho cos this guy does not just create monetary value for BTC via pizza purchase but was the first to convincingly use BTC as a medium of exchange which marked the beginning of its appreciative value.
Some may not have known that he also d scripted GPU mining code for Bitcoin, which he tested effectively and consequently used it to mine BTC at a higher hashrate.
His decision to buy papa John pizzas should be seen as an heroic gesture that has led to the success of BTC today cos as at that material time BTC neither had monetary value nor was it traded on any exchange.
Laszlo should be celebrated more as the BTC legend than an ordinary Pizza guy. I feel he should be accorded similar respect as Satoshi Nakamoto tho most CEXes are doing their best in rolling exciting events to comomerate Pizza day but I feel more should be in appreciating this great man.
Happy Pizza day 🍕 🍕 in advance guys 🎉.
Subscribe to:
Posts (Atom)