The Capital Cycle - Commoncog

This is Part 5 in a series on the expertise of capital in business. You may read the previous part here.

This is a companion discussion topic for the original entry at

Much of Marathon’s capital cycle analysis is built around established markets — and for good reason (they’re public markets investors, after all). I think it might be an interesting exercise to apply the capital cycle to a burgeoning market.

If we apply capital cycle analysis to, say, the generative AI space:

  1. We see a bunch of early startups growing as awareness of and demand for generative AI startups increase.
  2. With comparatively little barriers to entry, a swarm of copycat products and fast followers crowd into the space.
  3. Capital follows at around the same time, flowing into attractive AI products
  4. At some point, either demand falls off (due to a softening of the hype cycle), or supply outstrips demand.
  5. At which point price competition starts, eventually driving down returns below the cost of capital. (And at this point in our history, given US interest rates, the cost of capital is higher than what it was just two years ago).
  6. Capital flees the sector.
  7. Many startups get bought out, or shut down. The absolute trough of the capital cycle occurs not when market sentiment is the worst, but when the last marginal AI startup throws in the towel.
  8. The survivors are set up for supernormal returns in the future.

I don’t pretend to fully understand venture dynamics, but it strikes me that “how can you take advantage of this cycle” is not the right question to be asking here; the question is, “as a private markets investor, how can you invest at stage 6, given that capital is fleeing the sector and your LPs are likely going to be maximum bearish between stages 6 and 7”, and “how do you do venture deals, given that venture dealmaking is a game of access — i.e. you have to invest in earlier rounds (stage 3) in order to have follow-on rights in latter rounds (stage 6)?”

But as an outside observer, I’m fairly comfortable with the sequence of events, above. Any disagreements or tweaks?

Edited to add: and, if I was a bootstrapped operator in the space, it strikes me that the bits that would affect me is customer demand in Stages 4 through 7. In the sense that I’ll experience steep initial growth, think that it’s my genius when actually I’m just riding the rising wave that lifts all boats in this new sector, and then growth will hit a brick wall in Stage 4 and I’ll struggle to increase revenues … not knowing that perhaps there’s a supply demand mismatch that has to be worked out.

The tricky thing, of course, is that I won’t know if it’s my fault for sucking at marketing or product, or if it’s the capital cycle that’s at fault. But I guess that’s just what business is: making decisions in conditions of partial information.


How cyclical are education-creator businesses, specifically those for business/productivity information products?

What we’re seeing is that the market for cohort-based courses, like the ones that Tiago Forte, Khe Hy, and Ali Abdaal were running is contracting: all three of them do not offer their CBCs anymore and have transitioned to self-paced courses. So supply for CBCs is shrinking, the capital cycle is doing its thing.

But CBCs are only a packaging format for information products.
Information products are famous for their low barrier to entry, but Tiago, Khe, and Ali have built strong moats by creating unique brands.

That makes it hard for me to reason about the market itself – what’s the supply-side signal for an overall contracting business/productivity information product market?


I think the answer to this is ‘not very cyclical’ (at least, when compared with other industries), but I also think it’s clear that we’re at the tail end of one full turn of a capital cycle.

In my head, I think something like the following happened:

  1. Thanks to the pandemic, there was a surge in demand for virtual education + entertainment experiences, so cohort based courses became very lucrative very quickly.
  2. This attracted capital, which rushed into the space (see: the whole VC hoopla around the ‘creator economy’, the quintessential example of which was likely this HBR piece in 2020)
  3. At which point every creator started cohort based courses, and ‘creator economy startups’ spun up products that catered to cohort based courses and/or their communities.
  4. The pandemic ends, people get back to meat space, demand collapses for CBCs,
  5. There is an instant supply overhang due to the demand collapse.
  6. Capital flees the sector (leading to think pieces now, like this one)
  7. Companies and creators shut down their courses, or pivot, or consolidate their products. (We are here).
  8. The future: ??

I should note that the capital cycle is not a ‘law’ of nature, and Marathon spends a huge amount of time talking about how the pattern can play out differently depending on the specifics of the market, or the government, or globalisation, or technological disruption, or [insert novel and inconvenient factor]. This should not surprise us — at this point, we know that business is an ill-structured domain. I think it’s probably better to think of the capital cycle as “hey, here’s a pattern that seems to recur in markets … the same way that overpopulation and then population collapse seems to recur in natural ecosystems, albeit in slightly different forms.”

So what do I think would happen in the future? A few guesses:

  1. I think you’re right in observing that supply is not homogenous for CBCs: creators are not 1-for-1 replacements for each other. What we’re seeing is more a demand collapse for CBCs, to a level that is probably more ‘normal’ (Perhaps to some pre-pandemic level? But I don’t know what that is.) People just aren’t that into CBCs in normal times, especially if they can go play … pickleball or something.
  2. We may see some creators pivoting to CBCs for companies, though with the broader economy the way it is, I’m not sure what various companies L&D budgets is like. But I’d reckon demand there is higher than creator-based public-facing CBCs.
  3. Creator economy startups either throw in the towel or pivot to enterprise (I’m talking out of my ass here — I’m not privy to the shape of the game they’re playing so I don’t know if this will happen!)
  4. It’s not clear that consolidation in this market will lead to supernormal returns in the future; in this particular case, what we’re seeing is mostly a demand collapse due to a one-time special event, leading to supply overhang. But I think you’re mostly good if you were never geared up for that level of demand.
  5. It’s worth asking what the equivalent of a counter-cyclical move here is. I’ll leave @SAY to give his thoughts here, since he’s more believable than I am on creator economy businesses! :wink:

Note to self: I should probably catalog more cases of the capital cycle; that will be easier once Commoncog’s case library is up.


I have “insider information” on those 3 creators you mentioned so I can’t share my full thoughts too specifically here but ultimately I think the counter-cyclical move is to take the earnings from a creator business and use it to fund a different business model (like, say, SaaS).

A creator business’s audience is great for kickstarting the initial “true believer” users of a SaaS product. Money from an info product can fund a full development team for several years. You can quickly break out of the cold start problem if your courses and SaaS tool share a common audience and leapfrog most of the normal risk.

If a CBC creator started a SaaS at the peak of the pandemic, everyone else would look at them like they’re crazy. Why would anyone leave Courseland where money flows like water? Especially to put all their earnings into a seemingly much riskier, much more difficult venture? What does an internet writer even know about managing software developers, anyway? Just make more cohorts! Delegate those cohorts to a team! Increase supply!

But of course, post pandemic, jumping ship at the CBC peak would have looked like a genius move.


Another thing that’s struck me, as a result of thinking through the creator economy example (thanks @cortexfutura):

  • I’d like to revise my opinion: I don’t think info-products are cyclical. And it’s actually interesting to think about why.
  • What triggers a turn of the capital cycle is really there in step 1: capital flows to high return businesses. If businesses aren’t for some reason ‘suddenly’ high return, a capital cycle doesn’t even start!
  • This sounds stupidly obvious, so let me say why I think this is important: cyclical industries (by definition) contain within them the seeds of the next turn of the capital cycle. A cyclical industry like commodities or semiconductors are upstream of some level of (growing) base demand (everyone needs steel, minerals, chips, etc, and consume more over time) and therefore are set up for supernormal returns once there is enough consolidation on the supply side. But this is harder to reason about for new industries, or for industries where base demand is not guaranteed! So it’s not at all clear what demand is going to be like in the market for generative AI (since everything is still in flux, this being a new category), and it’s not at all clear if there is strong base demand for cohort based courses.
  • Another way of saying this is that, depending on the market, fads can produce one rotation of the capital cycle, but not more.

This is really interesting, and now that you point this out there’s ample evidence that they are doing now – Ali has been posting quite a bit about doing software and even released a Premiere Plugin I believe. Thomas Frank is going more and more towards building software on top of Notion (Flightlighter) and I’m sure there are many more examples.

Now the question for me, of course, as someone who a) did NOT make boatloads of money during the pandemic and thus can’t invest it into building SaaS solutions or similar and b) still managed to survive until the current trough of the capital cycle for creators in that niche is two-fold:

  1. What’s the next wave of supply from creators looking like?
  2. Presuming I’m able to capitalize on it in some fashion, how do I recognize the peak of that cycle. This is obviously the hard question.

My current hypothesis for 1) is that it’s content around AI for productivity, esp. in organizations. My indicator for this is the strong growth in AI-focused newsletters, which have the lowest barrier to entry. The wave of “buy my prompt library” offers has already crashed, imo (thanks to Elon killing the AI threads bros through an algorithm change – that’s certainly one of the external influences on the capital cycle @cedric), and now it’s likely going to move to higher production quality, higher barrier to entry courses.

The question is: where is the peak?
This actually poses a very thorny question, which is: how big is the AI wave itself?
That’s the question I’m not prepared to answer conclusively, my revealed preference/hypothesis is that it’s really fucking big and I should work more to get more out of it. :smiley:


Interesting discussion about CBCs, especially since I was considering doing one of those as my next project, and may reconsider in light of these inputs.

What I thought about while reading the article was the tech industry more generally, particularly this line:
“a small number of large players evolve from a situation of excess competition and suddenly exert ‘pricing discipline’. This occurs when the managers of the respective companies are all tired of war and are motivated by balance sheet repair.”

During COVID, the tech companies started paying outrageous amounts and giving employees every perk imaginable, including remote working conditions. Then over the last year, they pivoted away from that in unison, with layoffs and return to office initiatives. This shifted the balance of power from employees to employers dramatically because everybody knows now there are 100k people without jobs who will take your job if you don’t fall in line. With Google, Amazon, Meta and Microsoft all doing layoffs, this was a classic case of coordinated action without explicit coordination, completely changing the supply economics of chasing talent.

I can’t quite see the move to make to take advantage of being at the bottom of the talent capital cycle, but would love ideas. My sense is that this is the time to do favors for lots of folks, so when we return to “normal” and people have more negotiating leverage, they remember who helped them when things were tight. Curious what others think.

Cool article, @cedric, especially since I will never read that book! Took me til the weekend to find time to read the long article, though :wink:


As someone who has worked on „AI“ services for a few years now, my impression is that we’re already past the peak of the LLM hype. Which means the through of disappointment is coming up on the hype cycle. Supply might still be there a bit in the creator space, but is there really much demand?


This was a fun one from Byrne Hobart, the gist of it being: capital light businesses are amazing and desirable and high return and nice, so they attract capital to the sector, at which point the capital light businesses … stop being so capital light.

Software can still be a wonderful business with favorable economics, but capital will always flow into places where it can get a high return, and out of places where returns are low. There are traits of software companies that make them great, like high margins, the ability to pivot early, and the possibility of scale benefits like network effects and lock-in. But these traits must, over time, be offset by the competition to either start similar companies or to acquire customers for them. Absent excellent management, government protection, or extraordinary good luck, the industry’s economic balance sheet, as opposed to its accounting balance sheet, must end up looking like that of a capital-intensive industry like airlines.



Reader and investor (his exact title is Associate Vice President – Maritime, Air Cargo, and Energy Infrastructure at Stifel) Frank Galanti sent me this email in response to my note at the bottom of the Capital Cycle piece. I wrote:

(A caveat: I am actively looking for good examples of well-run cyclical businesses that are not conglomerates; I’ll update this takeaway if I find good counter examples. Unfortunately, pretty much everything I’ve read about good cyclicals seem to be this sort of ‘multiple businesses, portfolio approach’ type of company — with some moats. Is there an alternative model that works? I don’t know, but I’ll find out.)

Galanti’s response:

Maybe not 100p of what you were looking for per below, but the answer for cyclical businesses is low leverage and timing purchases and selling decisions correctly (while leveraging up to buy assets).

In maritime business, Kirby is a good idea example. They are a tug boat and barge operator in the US which is protected from international barges entering the market through the US Jones Act. Kirby was able to vertically integrate the maintenance it was doing for its tug boats to build a business that spit off cash to buy more barges. They are at something like 30-40% of all the tank (liquid) barges in the US, run low leverage, are investment grade, and lever up a bit to buy when the cycle is weak because they always have cash/cash flow.

On the international shipping side, public companies in shipping has gone to pure play sectors (containers Maersk/Hapag Lloyd, crude takers – dht/eurn, product tankers – stng/asc, dry bulk – SBLK, GNK, EGLE, LNG – FLNG, CoolCo, etc.). DHT is a good example of a company that does it correctly in a single ship sector. They have rules about how they operate (we will sell on the margins when asset prices do XX relative to the last 12 months, we will only buy second hand vessels, we will run below 40% net debt to capitalization, we will only have 1 type of ship – VLCC, etc.), have 2 CEOs, and are regarded as a high quality management team and get a premium valuation relative to peers.

Shipping magnates/privates diversify into other shipping subsectors that have different supply demand dynamics. So buy tankers now by selling your dry bulk assets (as an example). Running low leverage (sub 40% net debt to capitalization) on assets that can support 60-70% net debt to capitalization). Low overhead. But timing the cycles is what it is all about. Shipping returns get destroyed because shipyards are employment centers for China/Korea/Japan, so they are subsidized by the govts to keep operating, so anyone can buy a ship and they don’t have to pay for it until it is delivered (massive working capital needs/terrible business). But anytime something is going well, look at dry bulk going into 2008, orderbooks can go up to 50+% of vessels on the water and by the time everything gets delivered, demand growth usually can’t keep up.

I find Galanti’s explanation on DHT very illuminating — here is an example of a ‘single sector company’ doing well in an extremely cyclical industry. That the shipping magnates end up acting like the conglomerateurs they are is less surprising — when you have multiple businesses in your company, you have the luxury of focusing on those other businesses when times are tough for one particular sector. But I like Galanti’s overall observation, which is using “low leverage and timing purchases and selling decisions correctly (while leveraging up to buy assets).”

1 Like

I’ve been watching out for concept instantiations of the capital cycle in the shipping industry ever since reading Frank Galanti’s email, above. Couldn’t help but notice the news of Maersk cutting jobs yesterday:

As it is, as it always shall be.

1 Like

Adding some commentary about the Maersk Quarter results. Here is the commentary from an expert reporter

Maersk has released their Q3 accounts and in headline form the Ocean segment of their business was loss making, their logistics entities keep losing volume and the overall company is in the process of reducing staff by more than 10,000 people.

In their presentation they state that as part of restructuring the workforce of 110,000 people would be down to 103,000 at the end of Q3 and is expected to be reduced to below 100,000. It should be noted that some of these are not lay-offs but rather positions which do not get re-filled when they become vacant due to natural turnover of staff.

They clearly expect strong headwinds and are saying they will look at preserving cash which includes reviewing their share buy-back program as well as a reduction of the CAPEX by approx 1 Billion USD in 2023 and 2 Billion USD in 2024 compared to previous guidance.

The Maersk group as a whole did have a positive EBIT of 538 Million USD in Q3 2023, however the Ocean segment dipped marginally below zero with an EBIT of -27 Million USD.

For once Maersk’s Ocean business actually managed to gain marketshare in Q3. They saw year-on-year volume growth of 5% versus the CTS data which shows a year-on-year growth stagnating versus 2022.

In the same period the size of their operated fleet declined -3.1%.
Maersk saw freight rates decline -58% versus a global CTS rate decline of -62%.

Hence a comparatively positive development when compared to the global market data on these elements. But it was insufficient to keep the Ocean part of the company profitable.

The Logisics & Services part of the business remained profitable with an EBIT of 136 Million USD.

However the concerning element is the physical volumes they report. Supply chain management volumes measured in cbm declined -2.7%. Intermodal volumes measured in FFE declined -12.6%. Only airfreight saw in increase of +12.3% when measured in tons.

This time the cliff is too steep and they don’t see it coming back till FY2026.

Maersk over the last couple of years acquired multiple adjacent capability companies to provide all ancillary services required for container shipping.