AI in a bubble?
Bubble, bubble; toil and trouble: investments in Artificial General Intelligence (AGI) may be the next in the plot.
The trouble.
Asset bubbles are a natural feature of the investment markets. The challenge for the canny investor is to watch from the side-line or be part of the smart few – able to enter and exit early; not the overeager masses who are likely to suffer from the bubble’s deflation. This blog is NOT advice; rather just musings for anyone’s research in this matter. For the moment, one can set aside the ethical issue of the coming “Singularity” - for those interested in Legacies, a starting perspective can be found here.
History is littered with bubble examples (see Diagram 1). They have gone from being a rare occurrence of relatively small size to increasing in scale and frequency. They are not easy to spot in the early stages but are obvious when they deflate. Post-rationalisation? In the last 50 years, there have been at least 5 major bubbles and the peaks have steadily increased: for ‘Gold’ in the 1970s (at 500%) to the recent Digital Disruptors and Bitcoin (over 1,200%).
Diagram 1: Stock market bubbles through time: 1977-2022 and NASDAQ Index 1970-2022 (Source: BOFA, Bloomberg, Raktas analysis).
Turning to the underlying AGI technology of generative chatbots and the like (such as ChatGPT) there is scepticism. The Large Language Model (LLM) does not actually synthesize information. It is sequential responsive phrasing in real-time, based on probabilities. Therefore, the LLMs fail to provide thoughtful, analytical, or usually correct answers. Rather, the output is a stream of digital consciousness. Yet, other researchers claim that generative ChatGPT has passed a ‘Turing Test’; and is thus “aware”. It raises the question whether AGI’s underlying technology is a 3-D printing fizzle or ‘I-Robot’ in the making? AI has a function, as what? This uncertainty adds to the potential for promoting (investor) hype.
For the so-called ‘baby bubble’ of AGI, one witnessed a similar pattern in the past. The “dot-com” bubble of the Millennium may be a proxy for the sagacious investor. A 9-month balloon in the NASDAQ’s index took it to 9,500; only to see a spectacular collapse of 90%, down to 2,000 within a year. From that point, it took the index nearly 15 years to recover, driven by the FAANG-inspired “Disruptor” trend.
AGI brewing?
Today, the AGI bubble seems to be forming in classic style. A realistic idea being pumped by too much easy money chasing too few investable opportunities. The tulips have become LLMs.
There are no pure AGI plays; yet an increasingly large portion of the investment community want a piece of the action. That is what is driving the current growth (in share price and market capitalisations) of those Tech megacaps associated with AGI. Some analysts have estimated that these stocks are collectively overvalued by USD1.2 trillion. Look at Google (GOOG), with its AGI-induced undulations as well as; Microsoft (MSFT); Meta Platforms (META); Amazon; etc. All their share price increases are exceeding those of market indices by a significant margin. Then there are the facilitating companies; such as Nvidia (NVDA) with its share price up 30% since January 2023; and a further 20% in a day on Q1 quarterly results. Finally, the AGI-users, such as Buzzfeed (BZFD). Its share price climbed 95% in three days. based on the announcement of replacing journalists with ChatGPT functionality and its ilk.
It is an odd quirk that most hedge fund trading has been based on AI generated algorithms for at least two decades. The current clarion call is being institutionally led, with retail-interest gaining popular traction. Growing popular awareness and interest, is fuelling investments made; with no regard for fundamental analysis nor valuation. Such demand is leading to unrealistic expectations and encouraging every company to consider using the technology. FOMO rules!
On the other hand, there are those pundits who maintain that there is still a sensible investable ramp-up period for investors; before the AGI bubble truly swells beyond realistic metrics. In fact, it may not even be a bubble at all. Using the same dot.com parable, they base their arguments on comparing forward-looking P/E ratios from the 2000s and present (see Diagram 2).
Diagram 2: Comparison of “Hot Tech” forward P/Es - 2000 vs 2023 (Source: CNBC; Raktas analysis)
What is in a number?
One can observe that the speed of adoption is alarming. For context, to reach the 100 million user mark (mum) it took: the telephone 75years; the WWW 7 years; WhatsApp 3.5 years; Instagram 2.5 years; yet ChatGPT did it in 2 months! The crypto-craze saw over 10,000 coins minted within a handful of years.
A few encapsulating examples. Andreessen Horowitz’s recent USD 200 million funding round for Character.ai; gave it a USD 1.0 billion valuation - a company without product and revenues. Such frenzy has already hyped the likes of BigBear (BBAI); C3ai (AI) and Soundhound (SOUN)
What is that dome-ish feature on the water’s horizon; is that a swan? It can’t be a black one! Maybe a bubble?
Pied-piper on the riverbank.
Then there is fundamental research. The use-cases for AGI are proliferating faster than crypto coins. At the moment, there is little clarity as to what AGI is and what benefits/threats it poses. Yet, the likes of Goldman Sachs claim that AGI can enhance productivity by 30%. It has claimed that AGI will give the global GDP a (staggering) +7% annual boost! Others point to massive displacement in employment levels.
Some market analysts reckon that the current pure AGI market is worth USD 180 billion, due to increase to USD 2 trillion by 2030 (representing a 37% annual increase). They cite the fact that less than 25% of US corporations incorporate AI in their current business mix.
The assumption is that the AGI companies will grow into their TAMs; thus, the current valuations (as above) are justified. Some companies may indeed be taking genuine AI steps, such as Amazon (even Apple) to incorporate their own AI solutions with their own proprietary chips and software; which begets AMD and Broadcom. The index to look at is not the S&P but AI growth.
Where have the champions of WallStreetBets gone? Yet Silicon Valley is calling for regulation! If Central Banks ease their battle against the Dragon-called-Inflation, they may well stoke the bubble. The “Roaring 20s” all over again?
There is no disputing that digital Technology has exceeded expectations. But that fact has not necessarily seen ‘value’ transcend from Wall Street to Main Street. An economic conundrum that remains unanswered by the experts. Those original, originating, progenitor companies have not fared well.
Who under the age of 40 years if age has heard of AOL or Napster? A victim of this current bubble could well be Google itself. Long has it feared an end to its reign. An AGI-bubble-induced crash as well as the functional challenge posed by competitors’ LLM offerings may make its relevance redundant. Google risks going the way of GE, HP and IBM. Where are they now from their giddy heydays? No black swans, just white ones.
No Cassandra here
It is easy to criticise, and many stock prophecies are only worth a byte of capacity. Yet, how to be helpful? What to do?
AGI is a form of “frontier” investment. A time-tested approach may be best.
Be informed. Rely on proven analytical frameworks and common sense.
Be sceptical and ask hard questions for solid answers.
Seek out contrarian views to test one’s own.
Beware of “AI-washing”.
If it is ‘too good to be true’, then it isn’t.
Do the hard research.
Fire test assumptions.
Proceed, if at all, with caution.
Timing of entry is all.
Have a solid exit based on trigger points that are actively monitored.
If all of the preceding fail to satisfy, then ask Sam Altman what he is investing in and why? Otherwise, it may be prudent to take the funds and seek a better alternative, even cash. In such circumstances, buying a condo in Las Vegas and playing the casinos may be more rewarding in-the-round.
The advances of AGI and the possibility of the coming Singularity (even with a low probability) does pose some fundamental ethical, moral, social and economic issues that need to be addressed. “FOOM-day” may be equally hype inflated (aka Fast Onset Of Machines). Yet, the canny investor will try to discern a pattern and then placing steps carefully, pick a better path forward.
+
[This article is ChatGPT-free]
Justin Jenk enjoys finding and connecting dots