One Damned Thing After Another
Some thoughts on continuous overbuilding
In September 2021, at the height of the COVID-induced supply chain crisis, Nassim Taleb tweeted:
“I’ve seen gluts not followed by shortages, but I’ve never seen a shortage not followed by a glut.”
This couldn’t be truer over the last few years. It seems that every time there is a whiff of a shortage in some area of the economy, companies rush to overbuild. Time and again these shortages are followed by gluts and, inevitably, busts.
Just in the last few years we’ve seen mini booms and busts in:
Semiconductor chips
Vaccines
A land grab for engineering talent
Warehouse and distribution centers
GLP-1 weight loss drugs
The Metaverse buildout
Electric Vehicle investments
Environmental, Social & Governance initiatives
Most recently, data centers and power (this one is still squarely in boom phase)
Throughout history, humans have reliably swung from extreme to extreme, but it sure seems like the sums being devoted to these mini (or not so mini in some cases) booms is substantially increasing over time.
To be clear, some, maybe a lot, of the overbuilding that occurs is a good thing. Pfizer and Moderna basically saved the global economy with COVID vaccines, AI spending may end up changing the world, perhaps we’ll all be driving around in EVs within the next ten years, and so on. Just because some of the overbuilding is good or potentially good for society does not mean it’s good for the investors who participate in funding the overbuilding.
“Build it and they will come” seems to have been taken to a logical extreme time and again. Let’s look at just a few of these recent mini-booms and some of the fallout.
The Metaverse
Mark Zuckerburg changed Facebook’s name to Meta in 2021 as he pivoted the company to become a metaverse leader – whatever that means exactly. Since then, the company has plowed $77B+ into “reality labs”, it’s metaverse segment. The return on this huge spend has, thus far, been a loss of about $73B.
Investors quickly soured on Meta’s spending to build out the metaverse by voting with their wallets. In 2022, amidst peak concerns about the excessive spending in reality labs, Meta’s stock dropped 75%. Finally, Zuckerberg got the message and deemed 2023 as the “year of efficiency”. Spending across the company was temporarily reigned in, and Meta’s base business kicked into high gear. The stock has about 6x’d from its lows. Message received!
As we’ll see later, Meta looks to be caught up in another spending frenzy, but investors have not signaled that they’re ready for Meta to reverse course – yet.
Vaccines
Pfizer’s COVID vaccine was initially met with a torrent of demand. The company originally forecast that it would sell $15B worth of the vaccine in 2021. It actually sold $36B. Pfizer rushed to increase capacity. At the end of the year, Pfizer’s stock sat at an all time high.
As the pandemic waned people, perhaps predictably, stopped taking so much of the vaccine. Pfizer originally predicted around $25B of vaccine sales in 2023 and planned its operations accordingly. Actual vaccine demand came in at about half that.
Forecasting challenges are completely understandable, especially for Pfizer who was in the midst of supplying the vaccine for the worst pandemic in a century. The virus was changing rapidly, and no one could have been expected to predict what demand would be in a few weeks, let alone a few years. Still, the stock market doesn’t really care if making predictions for your business is hard.
What was Pfizer’s reward for coming out with a vaccine that basically re-opened the global economy and provided the business with tens of billions of dollars of unexpected (pre-COVID) revenue? A 57% stock price tumble from its peak in 2021 to today’s prices.
The stock currently sits near a 12-year low; how many people would have guessed that result in fall of 2020 when Pfizer announced that their vaccine was effective at guarding against COVID?
GLP-1s
The battle for market share in new GLP-1 weight loss drugs follows a similar trajectory as other booms, though with its own nuances.
Novo Nordisk had the early lead in GLP-1s. Novo pioneered these drugs with Wegovy (which piggy-backed off its diabetes drug, Ozempic) in early 2021. Sales boomed and Novo’s stock price followed suit. In the middle of last year, Novo Nordisk’s stock had the highest market value of any company in Europe, briefly eclipsing luxury empire LVMH.
Then the shortage hit. So many people were desperate for Novo’s new drug that they ran out of capacity. Eli Lilly and low-price generic competitors pounced on the opportunity and took massive share from Novo. Now, just as Novo scrambled to ramp capacity back up, pricing power is eroding. Both Lilly and Novo agreed to cut prices in exchange for Medicare and Medicaid coverage (and presumably to ward off cheaper competitors). Supply is set to continue to flood the market with Amgen and Pfizer introducing pills of their own.
The story is still not written on who will be the long-term winner of the GLP-1 race, but it doesn’t look great for Novo. Amidst the challenges of navigating and botching this mini-boom in which Novo had a clear pole position, its stock is down about 66% from its high last year. Once again – who would have predicted that just 18 months ago when everything looked rosy from the booming demand for its new magic shot?
Electric Vehicles
In January 2022 Ford announced an aggressive push into electric vehicles, highlighted by the release of its new electric Ford 150. In early 2022, amidst excitement over the company’s new EV lineup, the stock reached a multi-decade high.
Ford was piling in on EVs during a period of intense optimism from investors that EVs would quickly become the default purchase for car buyers. The company opened its checkbook to triple capacity for the electric F-150 and forecasted a 150,000 production target by 2023. Ford’s EV chief said at the time:
“We go from nothing to full-on production volume immediately. This is not a soft launch”
Fast forward to December 2025 and it turns out that excitement (and associated spending) was just a tad overdone. Ford sold about 23,000 electric F-150s through the first nine months of 2025, nowhere near its target of 150,000, and recently announced it would stop making the truck all together. Instead, Ford is going to focus on actually making money. CEO Jim Farley (who previously said his name was on the line after going all-in on EVs) said:
“Instead of plowing billions into the future knowing these large EVs will never make money, we are pivoting”.
Ford took a $19.5B write-down on its EV platform and its stock sits about 46% below where it peaked during the excitement of its announced EV ambitions in early 2022.
Data Centers
I won’t rehash the numbers or all of the details regarding the dollars being spent to build out data centers and AI capabilities. Matt wrote a great piece recently outlining some of the main drivers. I will remind everyone of the magnitude of the numbers involved. Current estimates are that AI related spending will hit $1.5 trillion in 2025 and ring in around $2 trillion in 2026. That’s trillion with a “T”. Fawkes Capital Management recently wrote a letter about the sustainability of this spend. They noted:
“If Big Tech and data centre operators collectively spend around $400 billion on AI infrastructure in 2026, then, by our estimate, at least $80 billion in annual net income would need to be generated to justify that investment. The hurdle is high because processors, which make up the bulk of capex, have a useful life of only about five years. Back-solving this requirement implies that something like 333 million paying users of ChatGPT – roughly the entire US population – would be needed to support such economics.
Today, the numbers fall drastically short. Only around 5% of users (about 20 million people) pay for ChatGPT, and both paid and non-paid user growth has begun to stall in recent months. OpenAI’s attempt to introduce advertising as a revenue stream has met fierce consumer resistance. And unlike Google, directing users to websites does not generate economic value for OpenAI. This raises the critical question: how will OpenAI, or any non-advertising-based AI provider, monetize its service at the scale needed?”
It’s very difficult for us to envision a scenario that the current AI buildout does not generate inferior returns for those laying the dollars out today. It doesn’t mean it’ll be a total waste for society, or even for some of the big spenders. If Google spends a couple hundred billion in capex to ward off potential AI competitors and ends up with a nice business unit at the end, it was probably worth it, though it still might be de-worsification compared to its core search business. But it won’t be a good outcome for all, and probably not for most.
With the dollars of the latest *potential* overbuilding dwarfing previous cycles, I wonder what it means for the players involved when the glut inevitably arrives and prices need to come down (for data center leases, chips, power, etc.). One thing I do know is that we aren’t interested in finding out, so we simply steer clear. There are easier, less risky, ways to make money, in our view.
Maybe the world’s largest spending frenzy won’t result in a bust. It’s possible, but it would also be a first.
An interesting exercise from an investor’s perspective is to envision a scenario where the current capex boom in AI infrastructure does not defy the laws of economics and eventually follows the same pattern as the other overbuilding scenarios discussed above. That means those that participated in the boom will likely see their stock prices drop 50-75%. This seems possible if not outright likely.
Some of the biggest spenders in AI are Microsoft, Amazon, Alphabet, Meta, Nvidia, and Oracle. Those six stocks represent about 27% of the S&P 500. If those stocks pull back 50-75% on average during the down leg of this capital cycle it would pull the index down 15-20%. That’s before considering the fear and contagion that may spread to similar sectors or trigger an actual recession in the economy. JP Morgan recently estimated that 41 AI-related companies make up 47% of the S&P 500’s market capitalization. The other 459 stocks represent 53% of the index. The S&P 500 has essentially become an artificial intelligence ETF, and we’ll see how that goes if the current spending spree proves overdone. It may work out just fine, and it may follow the patterns of history, time will tell.
It may seem outlandish to even suggest that these tech behemoths that have seemingly done nothing but go straight up could pull back by 50% or more. However, you don’t need to reach into the annals of history to find declines like this. Here are some recent drawdowns in these stocks:
Alphabet: 2022 (45%)
Microsoft: 2022 (36%)
Meta: 2022 (76%)
Nvidia: 2025 (35%), 2022 (66%)
Amazon: 2025 (31%), 2022 (58%)
Oracle: 2025 (45%), 2022 (41%)
All of these have occurred within the last few years and most of them occurred when the stocks were far less consequential to the overall index AND occurred during periods of relative prosperity for the businesses. It’ll be interesting to see what happens during a period of less than rosy financial performance, as could be the case if the current gargantuan spending in AI fails to generate commensurate profits. I’m not saying this will happen, but it would also go against historical base rates if something close to that didn’t happen.
British historian Arnold Toynbee was famous for explaining that “history is just one damned thing after another”. To us, that quote seems to apply to spending frenzies, especially in recent years. The examples I gave above are far from all the mini booms we’ve experienced recently; they were just what came to the top of my head when I decided to write this post.
Our main takeaway from watching these unfold is that it’s really hard to predict what demand is going to be in just a year or two in areas where demand is exploding. Whether it’s a novel vaccine or weight loss drug, new fangled technology, or the latest in transportation, it’s very hard to predict where these things will go. It gets dicey when companies have to spend a lot of money to try to get ahead of the potential demand that seems like it will continue in perpetuity. Demand in these areas can continue or it can collapse. Winners are next to impossible to call. The latest overbuilding cycle is no different.
None of this worries us because of what we own. We remain fully invested in companies whose future we feel great about. We do think it should worry some people, though.
We’re excited to watch the story unfold, especially because we won’t be betting on any specific outcome in order to earn our investment returns. When you own high quality, cheap, free cash flow generative businesses that tend to plod along regardless of how much tech giants are or are not spending on the latest capex frenzy, it allows you to observe the froth of the market from a distance, hoping for a chance to benefit for any wreckage.
Do you have a “stranded” 401(k) from a past job that is neglected and unmanaged? Eagle Point offer separately managed accounts to retail investors, and 401(k) rollovers are often a good fit for our long-term approach. If you would like to invest with Eagle Point Capital or connect with us, please email info@eaglepointcap.com.
Disclosure: The author, Eagle Point Capital, or their affiliates may own the securities discussed. This blog is for informational purposes only. Nothing should be construed as investment advice. Please read our Terms and Conditions for further details.