Chris Hondros/Hulton Archive via Getty Images
There’s an old joke on Wall Street, “this time it’s different.” The phrase invokes a sense of naivety about financial markets, becoming a running joke amongst financial professionals and amateurs alike. The history of Wall Street is a catalogue of speculative frenzies followed by dramatic fallouts, from the canal boom of the 1820s to the Reddit Rebellion. Dow 30,000 by 2008: Why It’s Different This Time is a crowd favorite.
Value investing legend Seth Klarman distilled this neoclassical cynicism to its core, as he warned of an impending asset price bubble in his 2014 letter to shareholders:
When the markets reverse, everything investors thought they knew will be turned upside down and inside out. ‘Buy the dips’ will be replaced with ‘what was I thinking?’ . . . Anyone who is poorly positioned and ill-prepared will find there’s a long way to fall. Few, if any, will escape unscathed.
–Seth Klarman, 2014
The S&P 500 has risen approximately 140% since this was written. More importantly, the Philadelphia Semiconductor Index (NASDAQ:SOX) is up approximately 550%. We will come back to the significance of this later.
In March of 2000, the Nasdaq 100 peaked after rising 365% from the start of 1998. The top of the market coincided with a peak in Cisco, briefly the world’s most valuable company. Microsoft peaked just before the start of the year, an omen of the bloodbath to come. Apple fell 70% over the course of 25 trading sessions from September to October. Lucent, which grew to become the sixth largest corporation by financing its own CLEC customers, climbed 112% between January and September, before narrowly avoiding bankruptcy roughly a year later.
And that’s usually where the story ends. Retail investors drove speculation in high-flying tech stocks. Even some staid fund manager types got swept into the mania, investing in Lucent and Cisco. The Dot-Coms burned up all their cash and went bankrupt. Greenspan raised rates. China joined the WTO and the country’s industrial base was ravaged until the economy could no longer be supported by middle management in 2008. The ‘knowledge economy‘ looked more like clever lobbying campaign for outsourcing.
A classic lesson in stock market speculation. The press and investment professionals alike have been drawing comparison between the outperformance of tech companies and the tech bubble ever since. See Bloomberg’s Silicon Valley Hears Echoes of 1999, published shortly after Seth Klarman’s letter.
Factorized comparison of Google to the S&P 500 (red)
But what happened after the tech bubble was something truly remarkable and little understood. Some of the most “over-hyped” companies lived up to the hype. Take for example Amazon, which was mocked with the infamous Barron’s “Amazon.bomb” article. Amazon crashed nearly 95% in the tech bubble, but is up approximately 220,000% since its IPO. Google’s post-crash IPO in 2004 was a dog, but it has rallied 5,300% since, becoming arguably the most powerful corporate institution on the planet.
Even some of the tech bubble’s most notorious names, such as Webvan, have seen their business models reincarnated elsewhere. Kroger is betting on robot-filled grocery stores and autonomous delivery. The first driverless car to cross the country did so in 1995, a decade before Tesla Autopilot or Google’s Waymo were conceived. Peppercoin failed. Bitcoin did not. Ethereum’s smart contracts were inspired by works from the 1990s.
Scenes from the tech boom and bubble era. The PBS series Computer Chronicles showcased various ideas, companies, and products. In the bottom left corner, an “internet terminal” that connects to MSN. In 1999, the internet was often viewed as a threat to Microsoft. This terminal was an attempt to create a Microsoft subscription model.
While some ideas of this era were poorly guided and destined to fail, investors were generally more early than wrong. For as much mental real estate as it has in the psyche of the markets, the tech bubble was but a flash in the pan relative to the internet’s ever-accelerating transformation of society. Note that in 1999, many viewed the internet as a threat to Microsoft. Microsoft generated $82B in operational cashflow LTM, enough to buy the entire company in 1996, buy Google in 2005, or buy Palantir (NYSE:PLTR) twice today.
The neoclassical premise of value investing seeks immutable rules through reason and objectivity. This pursuit may be a refreshing break from the nihilism of NFTs “worth” millions of dollars, but aside from the obvious, it fails to locate them in the broader context of technology. Value investing often dismisses companies as outside the appropriate level of discipline. In this section we will adapt this approach to the context of technology investments.
An excerpt from the book Race Against the Machine
Erik Brynjolfsson, Andrew McAfee
The first thing we must understand is the transformative effect that technology has had on civilization (World GDP is already nearly double what the graph above shows). Computers, which first entered commercial use with the UNIVAC in the 1954, have allowed us to quantify information with ever-increasing sophistication. Value investing, which was pioneered by Graham and Dodd then perfected by Warren Buffett, has been primarily developed at the tail-end of the industrial revolution.
The MOSFET transistor (1959), integrated circuit (1960s), and microprocessor (1971) wrought mind-bending leaps in performance over early vacuum-tube computers. The ever-increasing density of transistors on a chip has enabled computer performance to double approximately every two years. This is known as Moore’s Law.
Most investors are familiar with the concept of Moore’s Law, yet few understand what is happening today. Even fewer understand the relationship between hardware and software.
Today transistor scaling is slowing down, yet we are simultaneously at an inflection point that will accelerate the progress of semiconductors. Some technologies, such as ASML’s EUV lithography (NASDAQ:ASML), innovative FET features and new approaches to metallization extend Moore’s Law. Others, such as 3D-stacking, heterogenous integration, programmable memory, tiles, and chiplets are picking up where it leaves off.
AI chips typically provide a 10-1,000x improvement in efficiency and speed relative to CPUs, with GPUs and FGPAs on the lower end and ASICs higher. An AI chip 1,000x as efficient as a CPU for a given node provides an improvement equivalent to 26 years of [Moore’s Law-driven] CPU improvements.
-CSET, April 2020
Furthermore, acceleration architectures (referred to as “AI chips” above) are pushing performance improvements at a blistering speed. Chips are shifting from general purpose configurations to more application-specific designs with dramatic results. Chiplets will allow many different application-specific subsets to be integrated into a single chip.
As AI plays an increasing roll in the design of these semiconductor architectures, the CEO of Synopsys believes that performance will improve an addition 1,000x over the next decade. This will create a cycle of AI designing better chips that will, in-turn, create more powerful AI capable of designing yet even more powerful AI chips.
It also seems that software progresses at least as fast as hardware does, at least in some domains. Computer scientist Martin Grötschel analyzed the speed with which a standard optimization problem could be solved over the period 1988-2003. He documented a 43 millionfold improvement, which he broke down into two factors: faster processors and better algorithms embedded in software. Processor speeds improved by a factor of 1,000, but these gains were dwarfed by the algorithms, which got 43,000 times better over the same period.
-Andrew McAfee & Erik Brynjolfsson, Race Against the Machine
There is also a non-linear relationship between hardware and software. Hardware improves at an exponential rate, yet software improves at an exponential rate that is orders of magnitude higher than hardware. This makes software businesses incredibly powerful. They can not only capture this dynamic, but can scale quickly at low (in some cases almost zero) marginal cost. This dynamic creates economies of scale unlike anything else, often enhanced further by the ability to collect more data.
Though not strictly correlated, better hardware has enabled the creation of larger and more sophisticated AI models. We will likely see a 100T parameter model, something that seemed like fantasy until recently, before the end of 2022.
Nvidia (updated to show 1T parameter model from Google and 10T parameter model from Alibaba)
This is far from the realm of academic discussion; we are seeing exponential increases in the size and sophistication of AI models the real world. The power of AI will extend significantly beyond a subset of software applications. For example, Google developed an AI model that calculated the structure of nearly every protein expressed by the human genome. This will accelerate R&D in the medical field. It has been called “one of the most significant contributions that AI has made to the advancement of science“.
The advent of AI obliges us to confront whether there is a form of logic that humans have not achieved or cannot achieve, exploring aspects of reality we have never known and may never directly know.
The AI revolution will occur more quickly than most humans expect.
At every turn, humans will have three primary options: confining AI, partnering with it, or deferring to it.
-The Age of AI: Henry Kissinger, Eric Schmidt, & Daniel Huttenlocher
Yet matters are complicated by the fact that intrinsic value does not scale with model size or hardware performance. AI models typically have very little economic value until they are successfully trained, going from useless to disruptive almost overnight. Risk and reward are skewed to the peripheries. People say ‘data is the new oil’ offhandedly; few consider how radical a departure from traditional investment norms this will append.
Even the model of a joint-stock corporation itself will have to compete with blockchain-based and software-defined institutions known as DAOs, which themselves may become platforms for AI among other possibilities.
With the possible exception of a few incredibly well positioned (and/or entrenched) platforms, the inherent risks challenge the concept of ‘lifelong buy and holds popular’ among value investors. It is difficult to find “tech companies” that have true moats, but we will give examples in the next section.
Even the present rate of growth will produce impressive results if maintained for a moderately long time. If the world economy continues to grow at the same pace it has over the last fifty years, then the world will be some 4.8 times richer by 2050 and about 34 times richer by 2100 than it is today.
Yet the prospect of continuing on a steady exponential growth path pales in comparison to what would happen if the world were to experience another step change in the rate of growth comparable to in magnitude to those associated with the Agricultural Revolution and the Industrial Revolution…
…If another such transition to a different growth mode were to occur, and it were of similar magnitude to the pervious two, it would result in a new growth regime in which the world economy would double in size about every two weeks.
Such a growth rate seems fantastic by current lights. Observers in earlier epochs might have found it equally preposterous to suppose that the world economy would one day double several times within a single lifespan. Yet that is the extraordinary condition we now take to be ordinary.
-Andrew McAfee & Erik Brynjolfsson, Race Against the Machine
How exactly does one value a company in this context? It is, admittedly, extremely difficult. Boeing’s planes will not improve 1,000x over the next decade. Nor will Exxon’s oil or Darden’s food. A rotation to companies picked out of the bargain bucket, largely disadvantaged in the epoch of this disruption, is likely to be another short-lived fad.
A tell-tale sign that this “rotation to value” is a fad and not real economic trend derived from a change in interest rate policy is the fact that companies such as Delta Apparel (NYSE:DLA) are trading within range of all-time-highs. This anticipated hiking cycle might be enough to make the high-debt capital structure unsustainable, burning up its $9.4M cash reserve quickly. Perhaps owning a ‘value’ play that could get wiped out if it refinances its debt burden at a higher interest rate is the Bored Ape Yacht Club NFT of the boomer generation.
In this context, it seems difficult to believe that a “rotation out of tech” could be sustainable amid such rapid innovation and growth. In this section, we will look at the empirical evidence and attempt to find our bearings.
In contrast to academic dogma about efficient markets, the crux of the matter is that investors (both institutional and retail) struggle to do adequate due diligence. The trash needed to be taken out, but Wall Street has called for a complete portfolio eviction that has left valuable heirlooms on the street curb.
Factorized comparison of Palantir (blue) and Peloton (red)
This is to say, they have bought and sold indiscriminately. Take for example Palantir and Peloton, both down dramatically from their all-time-highs. Peloton (NASDAQ:PTON) makes high-end exercise bikes and burned $1.1B in operational cashflow LTM, less than it has on the balance sheet. Palantir (PLTR) owns a generalizable platform capable of transforming companies into ‘software production entities’ that incorporate data into almost every aspect of their operations (which can be installed almost overnight). It has $2.5B in cash, minimal debt, and $222M in operational cashflow LTM.
Another example is Core Scientific (NASDAQ:CORZ), the largest bitcoin mining operation in North America, backed by BlackRock. The rotation FOMO has been so comprehensive that you might have missed its de-SPAC debut. It is valued less than Riot Blockchain (RIOT), despite dwarfing it (down 22% on Friday alone). Core Scientific’s former CEO was the COO of Microsoft, who played a key role in recruiting current talent. Riot Blockchain’s former CEO was charged by the SEC for running a “classic pump-and-dump scheme.” Core Scientific has approximately $200M worth of bitcoin, in addition to $190M in net cash proceeds from its recent SPAC transaction.
First Solar stock price with the TAN ETF price overlaid (red)
Looking for something a bit more tangible than software and digital assets? Try First Solar (NASDAQ:FSLR). First Solar is the only company outside of Asia that can produce solar panels at significant scale. It does so using a technology that no one else has; First Solar can literally print a solar panel onto a sheet of glass. It has $1.9B in cash, minimal debt, and is sold out for several quarters into the future. Skyrocketing prices for polysilicon and the potential reduction of California’s rooftop subsidies has weighed on the solar sector. First Solar does not use polysilicon or produce rooftop panels.
So what does deserve to be thrown out with the trash? How about: Vuzix (NASDAQ:VUZI), which has been promising that huge growth is ‘just around the corner‘ for over two decades. Why not: Electrameccanica (NASDAQ:SOLO), which accused Seeking Alpha of being ‘an Israeli conspiracy‘, used company funds to buy Tesla’s and a Lambo, and has still not made any major deliveries despite claiming mass production would begin in 2017. Try: ViaSat (NASDAQ:VSAT), which is under existential threat form SpaceX. Or: (NBEV), which has an exciting business model that consists of rolling up unprofitable beverage companies.
A recent interview with Morgan Stanley’s Andrew Slimmon summarizes this moment perfectly. After warning not to buy the “growth-stock dip” because “once the fever breaks it’s done for quite a while”, Silmmon says he ‘believes in technology long term’ and says he really likes Microsoft (NASDAQ:MSFT) and Alphabet (NASDAQ:GOOGL).
One feature of markets is that no one is ever held responsible for falsely predicting a catastrophic crash, but anyone who is overly bullish is swiftly ridiculed. Fund managers have spent the better part of a decade predicting an epic crash with total impunity. Ark Invest’s Cathie Wood (ARKK)(ARKW)(ARKG)(ARKQ)(ARKF)(ARKX) is being publicly humiliated despite having a five-year record that towers over Warren Buffett (BRK.A).
In these matters, as often in our culture, it is far, far better to be wrong in a respectable way than to be right for the wrong reasons.
-John Kenneth Galbraith, The Great Crash 1929 (published 1954)
Hopefully this analysis has been sufficient to conjecture that concepts of value are highly arbitrary in today’s context, particularly where augmented by the strange risk/reward distribution of AI, the unusual economies of scale of software, and the exponential effects of better hardware. Amidst the chiding about valuations and fear-porn about a ‘big crash’, it is easy to forget why technology-leveraged growth companies (and furthermore cryptocurrencies) became so valuable in the first place. We look to the horizon with confidence until there is volatility, at which time horizons shrink to an immediate assessment of ‘here and now’ risks.
(1/3/22) – (1/21/22)(?)
Sell-offs in the NASDAQ Composite Index over the last decade
Even the CME Group (CME), which makes more money if investors rotate into commodities, has stoked the flames with an article titled: The Great Global Portfolio Rebalancing, which was syndicated on Seeking Alpha. The article points out that U.S. household assets have outstripped nominal GDP without mentioning that the fact that corporate profits have also grown much faster than nominal GDP, while real median household income has struggled to exceed 1990’s levels. In other words, technology and globalization have dramatically changed the US economy.
If this is yet another minor episode, it will be remembered as just one of many volatile shakeouts, separating the speculators from the investors. If it is a more serious affair, destroying the value of Palantirs and Pelotons alike, it may be ultimately remembered as the last great drawdown before the exponential dynamics of hardware and software catalyze a truly biblical transformation of business, and furthermore society as a whole. Our bet is on the first, and we are buying the dip here.
This article was written by
Disclosure: I/we have a beneficial long position in the shares of NVDA, GOOGL, PLTR, MSFT, CORZ, FSLR either through stock ownership, options, or other derivatives. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.