In the first part of this article I outlined how history repeats itself for investors as capital chases growth. The excitement of those heady days often turns out to be illusionary as outsized returns lead to steep losses.

As a reminder the capital cycle has four distinct phases:   

Phase 1: The advent of a compelling narrative

Phase 2: Irrational exuberance

Phase 3: Bust

Phase 4: Deep pessimism and the pathway to recovery

I provided two examples from history that showed the danger of getting caught up in the hype. Both the .com bubble and bust and the more recent buy now pay later craze show how the emotional aspects of investing can cause the market to collectively lose touch with reality.

Anyone can identify a bubble after it has popped. And while history offers lessons the key is to apply those lessons to future bubbles. That is how we can avoid them and the subsequent loses. I’ve identified two cases where investors may get burned as we continue to move through the capital cycle.

Artificial intelligence

If you haven’t been living under a rock the last few years you’ve probably heard about AI. This is bit of difficult narrative to put together from an investment standpoint. But that isn’t necessary. The narrative for the technology is clear. AI is a magic wand that will transform the way we do everything. How is this for a narrative – according to one article I read AI will “shape the future of humanity across nearly every industry.”

Who wouldn’t want a piece of that?

AI reminds me a bit of the .com bubble which I wrote about in the first part of this article. There is a new technology that undoubtably will have a profound impact on life. What that impact will be is not entirely clear yet. But this ambiguity hasn’t stopped investors from going wild for anything and everything that is remotely associated with AI.

In the first article I outlined five lessons from the .com boom and bust. As a reminder they are as follows:

  • Lesson one: While emerging technology can be exciting it is also incredibly disruptive to the existing order.
  • Lesson two: New technology generally costs a lot to roll out and commercialise and while capitalism is designed to funnel investment to these opportunities it also tends to overly fund certain opportunities through irrational exuberance from investors.
  • Lesson three: Even when an investor manages to pick a winner the price paid for the company matters when valuations skyrocket.
  • Lesson four: The ultimate beneficiaries are not necessarily who investor’s initially suspect.
  • Lesson five: Execution matters. A lack of capital requires disciplined decision making on what projects to fund. A period of limitless capital removes all that discipline.

I believe we are early in the AI cycle and while there is a potential for a bubble to form it may take some time. Nevertheless, it is worth keeping these lessons in mind as we watch developments. There are some early signs that the lessons from .com are applicable.

Lesson one: While emerging technology can be exciting it is also incredibly disruptive to the existing order.

Many investors may be ignoring the signs of the disruptive impact of AI. Meta META and Alphabet GOOGL have been classified as AI companies. And they are spending a great deal of money on AI with Meta projected to spend at least $37 billion US this year on AI and even more next year. Alphabet splashed $145 million US per day on AI infrastructure during the second quarter.

The reason technology companies are attractive is because they are capital light and generate ample cash flow for shareholders. This level of spending is not indicative of a capital light business. The driver of this spending is likely fear that the moats they have built are at risk from AI.

Before AI we had big data. Big data involves the collection of data and the analysis of that data. The companies that were best at the analysis made lots of money. The insights they derived from the data could be used to sell products and services or to sell advertising to help other companies sell their products and services.

Anyone can collect data. It is the analysis that is the competitive advantage. In both social media and search Meta and Alphabet have captured huge market share. They have done this because Meta has built algorithms that can continuously feed users with content based on their history. Alphabet has provided users with search results that answer their questions. If AI can do this better than competitors may have an opening. There is nothing wrong with investing to protect market share and a competitive advantage. It just has to work.      

Lesson two: New technology generally costs a lot to roll out and commercialise and while capitalism is designed to funnel investment to these opportunities it also tends to overly fund certain opportunities through irrational exuberance from investors.

One obvious investment opportunity is the infrastructure needed to support the build out of AI. Largely this has centred on the microchips and data centres that enable the computing capacity needed for AI.

The perfect example is Nvidia NVDA. The graphics processing units (“GPU”) that Nvidia traditionally produced for the gaming industry handle the parallel processing workloads needed for video games. Traditional chips called central processing units (“CPU”) sequentially process workloads. AI requires GPUs. Other chip companies have not ceded AI to Nvidia. Competition and increased capacity are in the works. Will too much capacity be created? Will competition lower prices? We will see. But if history is a guide the early success from Nvidia will not go unchallenged.  

AI also requires significant processing power and storage to enable companies to run AI models. Everyone from Microsoft and Amazon AMZN to Telstra TLS and Goodman Group GMG locally are building out capabilities and capacity in this space.

This is why it makes intuitive sense to invest in this AI infrastructure. If AI does transform the world and is used by more and more individuals and companies it doesn’t really matter which companies are disrupted and which are the winners as long as more infrastructure is needed.

When we examine this infrastructure play through the lens of the capital cycle the question becomes not if the infrastructure is needed. The question is if too much supply will be created which will create ruinous competition. We saw this during the .com boom as Worldcom and Global Crossing created so much supply of broadband that prices plunged. They both went out of business. We saw this during the railroad bubble in the 1800s in the US and England as so much track was laid that most railroad companies went of business. A sure fire opportunity can collapse if too much capacity is created.

Lesson three: Even when an investor manages to pick a winner the price paid for the company matters when valuations skyrocket.

The big AI question is if investors are paying rational prices. In the first part of this article I shared the example of Microsoft MSFT and Cisco CSCO to illustrate how overly optimistic investors can push share prices too high. In that case it took decades to top the highs of the late 90s / early 2000s. It took Microsoft 16 years to match .com highs. Cisco is at 24 years and counting.

Valuations are elevated for some of the AI related companies. Nvidia is trading at 75 times earnings. Microsoft at 35 times earnings. Goodman Group is trading at 4 times net tangible assets with a forward yield of less than 1%. To justify these valuation levels requires significant growth. Perhaps it will eventuate but it puts a good deal of pressure on these companies to deliver. With the exception of Goodman our analysts don’t think these are outrageous valuations but it is worth tracking going forward.

Lesson four: The ultimate beneficiaries are not necessarily who investor’s initially suspect.

We will have to wait and see if this lesson is applicable. This is the infancy of AI. As much as we like to play with ChatGPT we haven’t seen the true impacts of AI yet. The question is if the beneficiaries will be the infrastructure plays, the companies that provide AI technology or the companies that use it to become more efficient. If AI becomes a commodity it is most likely the later. More to come on this front.

Lesson five: Execution matters. A lack of capital requires disciplined decision making on what projects to fund. A period of limitless capital removes all that discipline.

As Warren Buffett said it is only when the tide goes out that we see who is swimming naked. All I know is that anything AI related is getting funded generously. Some of these projects and companies will work out. Some won’t.

Private credit

Private credit funds raise money from investors and lend it directly to companies. The narrative to these investors is straightforward – and really enticing.

The summary is that banks have been forced to pull back on lending due to post-GFC regulatory requirements which mandate higher capital levels. Private credit has filled this void and offers investors double digit returns that are reliable and stable.

That is a proposition that is hard to resist.

As a testament to the compelling nature of this narrative we need to look no further than asset flows into private credit. Institutional allocation to private credit has increased in the last decade from around under $600 billion US to over $1.6 trillion US according to Pitchbook. Over $506 billion of this total remains as ‘dry powder’ which means it is yet to be lent.

Private credit AUM

Australia’s largest super funds have been among these investors. Cbus and HostPlus’s default MySuper options have grown their unlisted credit allocation to around 7% at the time of writing. AustralianSuper has around 4.5% allocated but has publicly stated its intention to ramp this up. Private credit is also being pitched at retail investors through funds and ETFs.

This is a prime example of the potential for problems arising from too much cash flowing into an asset class. There are a lot of incentives for the private credit funds to lend this cash out. Afterall this is how they make money. There is no limit to the companies that want to borrow money. The ambitions of lending a huge pot of money and a near limitless supply of companies that want it has the potential to cross the line where lending starts to go to companies that are less likely to pay it back.

That has happened throughout history in the banking industry. And it is why the banking industry is heavily regulated. The wisdom of letting unregulated private credit to enter the market will be tested at some point. Caveat emptor.

There are some troubling signs that things are starting to unravel in private credit. The AFR has reported that ASIC has set-up a task force to look at how private credit firms are valuing assets. Two weeks ago the AFR reported that a quarter of loans in private credit provider Millbrook’s main fund are overdue. And this week the AFR report that AustralianSuper lost $1.1 billion in equity and private credit loans to US video training provider Pluralsight.

There is no transparency into what companies have borrowed money from private credit lenders. I don’t have a crystal ball. I am in the dark just like the millions of Australians who have invested in private credit through their super funds and directly through funds and ETFs. All I know is that there are always more people that want to borrow money than will pay it back. And lending standards tend to drop when there is too much money to lend. As Johnny Cash sang, I hear the train a-comin', it's rolling 'round the bend…..

Final thoughts

I started my first article on the capital cycle with a reminder that those that ignore history are doomed to repeat it. The lure of riches from shiny new investment opportunities often end poorly. In investing boring is sometimes better.

AI looks like the next revolutionary technology. That doesn’t mean the average investor will get rich. Especially investors that are deciding to take the plunge today.

Private credit is nothing new. The first coins were struck in China sometime around 640 BCE. Minutes later somebody probably borrowed a couple of them. This seems like a classic example of a banking crisis with new packaging. There have been so many banking crises it is hard to fathom that we are going down this road again. I could be wrong, but I see no way this ends well.

Whatever happens with AI and private credit will offer lessons for investors. Lessons that will likely be ignored. But time and time again too much capital flowing into any type of investment leads to pain.

I want to know what you think of AI, private credit and any other investment where too much capital is flowing. Write me at [email protected]   

Related articles

Get more Morningstar insights in your inbox