Extrapolating META's Excess Compute Narrative
Price Action Week-in-Review for 6/29/26 - 7/4/26!
Welcome to this week’s Price Action Week-in-Review for 6/29/26 - 7/4/26!
This is where we analyze the Price Action over the past week across different asset classes to identify outliers. From the outliers we develop an explanation of what is driving the price action. Then we derive market lessons to help us become better traders.
If you are new here and would like to learn how to understand the Price Action Week-in-Review, click here for an Introductory Primer
Let’s get into it:
Executive Summary
META announced they will start selling AI compute power, challenging the AI Narrative that compute is short in supply, necessitating heavy capital investment in datacenters
A weak jobs numbers on Thursday lowered concerns that the Fed may hike interest rates in September
Analyzing META’s announcements is more nuanced than just a binary decision of “Did the bubble popped or not?”
Disclaimer: The information provided on this site is for informational purposes only and should not be considered financial advice. All investments involve risk, including the loss of principal. Please consult a qualified financial advisor before making any investment decisions.
Indexes
Price Action for Indexes were interesting this past week. The Nasdaq 100 performed the worst for the day and, surprisingly, the Dow Equal Weight performed the best. This suggests that capital is now flowing from the highly valued AI names into blue chip DOW names in a “Great Rotation” towards safety.
There are a few reasons for this including META’s surprising announcement that they will transition to selling excess compute as well as the jobs report on Thursday that will be discussed below.
Industries
Again, technology performed the worst with both Technology and Telecom with > 2 STD declines for the day led by META’s Announcement of transitioning to selling excess AI compute power capacity. This fundamentally challenges the AI data center narrative that current capacity is limited and the only way to satisfy AI demand is to build more data Centers. When META as one of the major hyperscalers starts selling “excess” compute, the market is wondering what’s really going on in the AI industry. I go further in depth in the Market Lessons section of this article.
META’s announcement is the main story for the week. However, we shouldn’t ignore how Consumer Staples, Health Care and Utilities performed well for the day after the jobs report as part of the risk-off rotation. Healthcare has done well for the month, week, and day timeframes while underperforming YTD. Therefore, it is worth investigating further if there are any healthcare growth names that markets can rush into. UNH and CVS come to mind as names worth investigating further.
Bonds
Bonds did poorly early in the past week as part of markets pricing more than a 60% chance of a Fed rate hike. Thursday’s jobs report had Nonfarm payrolls rising by 57,000 well below the 115,000 consensus and slower than the downwardly revised 129,000 in May with April and May estimates revised lower by 74,000. After the report, Fed expectations of a September rate hike fell from 64% to roughly 50%. This was demonstrated in the 1 day upturn of bond prices but did not overshadow the damage done earlier in the week.
Commodities
Oil continues its downward descent as Hormuz flows surpassed 10 million barrels per day, the UAE restored exports to over 3.9 million barrels daily, and Iranian exports jumped above 40 million barrels following the lifting of the US naval blockade. Positive progress has been made between the US and Iran which is reflected in the price action.
Silver rallied aggressively, contrary to the downturn in tech, as part of the unwinding of Fed rate-hike bets as well as beliefs that inflation has tempered. Gold participated in this boost, albeit, to a lesser extent. However, Copper did not participate as much and was relatively flat for the week compared to the other metals.
Bitcoin also bounced off deeply oversold levels rising 2.56% following the June jobs report miss, bouncing bitcoin back above $60,000.
Market Lesson for the Week
The most important development is META’s transition to sell excess AI compute power capacity. This is a very big deal that needs to be explained in greater detail so you can make sound decisions about your portfolio moving forward.
META’s Position in the Big 5 Hyperscalers
The 5 big hyperscalers are GOOG, AMZN, MSFT, META, and ORCL. META was the weakest of the bunch because it focused on developing AI on its own for the purposes related to advertising efficiencies rather than selling computing power like the others. The market was not convinced that advertising alone warranted such high capital investment and preferred the business model of GOOG and AMZN.
The AI Thesis
The Fundamental AI thesis is based on the belief that there is a large demand for AI services and, therefore, AI compute. Consequently, this demand can only be satisfied with large capital investment in AI Data Centers. This has been the prevailing narrative to justify large investment from Private Equity, large issuance of stock, borrowing large sums of money, and significant declines in operating cash flows. Thus, if the investment was made, the customer would be ready with a fistfull of cash due to how amazing AI is (and will become). This explains the significant bottle necks and shortages previously in Silver and now in Memory.
The AI Thesis Challenged
META’s announcement challenges the AI Thesis because META should not have excess capacity if (1) AI is so amazing that all processing power needs to be used for AI; (2) META is experiencing benefits from AI to justify the cost of investing in building data centers. The fact that META has excess capacity suggests that either AI is not so amazing and further investment in building data centers is not necessary.
This could certainly be a fault of META alone. Perhaps they are AI losers, their models suck, and their implementation of AI is woefully below the competition to the level of actual incompetence. However, that’s a very extreme view that would need more evidence to corroborate.
It’s more likely that something is going on and many people are calling META’s announcement to be the beginning of the AI bubble bursting.
The Trade
It’s very tempting to think in binary terms when trading. Prices either go up or down. What more do you care about?
I’ve come to appreciate reflecting on the different players to understand how price action could play out. It’s more than just sellers and buyers. Allow me to demonstrate.
The Relevant Participants:
People holding AI Equities and are long the AI Narrative
People not yet holding AI Equities (or holding less than what they wish to have), concerned about the potential over valuation
People Shorting or wanting to short AI Equities aggressively
People who are undecided
People in META transitioning their capital expenditures to selling AI compute power
People finding new competition in this space like CRWV and NBIS
The Hyperscalers who are planning to spend billions in further Data Center buildouts
Data Center Component manufacturers (e.g. NVDA, MU, AVGO, CRDO, and more!)
AI Models - Open AI, Anthropic with a lot of debt
Private Equity with a lot of debt
And more!!!
So how do these connect?
It is in no one’s interest to announce the death of AI, even if it is potentially true. META was the outlier since they did not have a cloud service. Now that they are potentially offering to rent out compute, they are potentially in line with the other hyperscalers. However, META has not produced a convincing narrative why the market should not interpret this shift as a sign that AI compute demand is not as high as projected.
NVDA (and other component companies), AI Models, and Private Equity backing nearly the entire industry all have an incentive to spin this in a positive way. They are likely chatting with Zuckerberg right now to sort out what the store is going to be over the wonderful July 4th weekend. At least, that’s what I would be doing if I were any of those actors. “DUDE! We’ll stop making fun of you for your pathetic use of AI just for advertising. Just don’t screw this up for the rest of us! DOOOODDDDD!!!”
What could they say?
They could focus on how AI is more than just for advertising and that META is not structured to take advantage of that. So diversification of their compute is a good move on a business level (golf clap)
But they need to then answer for the significant revenue decline CRWV and NBIS will experience when their contracts with META are no longer needed.
Solution: AI Models could announce taking up all of their compute from both CRWV and NBIS because they created a brand new model that needs it asap. And with more supply, compute is cheaper and we’re THAT much closer to AGI. Woohoo!! THis is just a small speed bump along the way.
Private Equity breathes a sign of relief.
Some money will be shuffled to META just to keep them happy. And the AI narrative has been saved!!!
The above is not a prediction but just an example of how things could play out. I write the above to illustrate that the multiple players makes the assessment far more complex than “bubble popped” vs “bubble bigger”. Each of these players have many levers that they can pull to boost the AI narrative. But there are also events that could further deteriorate the AI narrative. When it deteriorates further, the levers the players can pull become less effective.
That is why the psychology of the market participants is so important. Namely
People holding AI Equities and are long the AI Narrative
People not yet holding AI Equities (or holding less than what they wish to have), concerned about the potential over valuation
People Shorting or wanting to short AI Equities aggressively
People who are undecided
These people are not a monolith and they will have a wide and diversified series of opinions about the subject. These opinions line in a spectrum that is easy to outline as either strong bull or bear cases. However, the trade is acknowledging the space in between and keeping track of the information as price action reveals itself. Easier said than done, of course.
META signaling excess capacity is not good no matter how you slice it. I expect prices to further decline until we hear from proponents of AI pointing to other data points that support the AI narrative. When those positive AI narratives result in further negative price action, keep selling. When negative commentaries about AI begin to hit mainstream media with doom scenarios, going into detail about the level of debt, and how challenging it will be to service this debt (a fact that has always been known by the markets since day 1), then watch for the selling to stop.
When the selling stops after continued negative news (especially negative main stream “analysis” news that begins to pervade every corner of the internet), a couple of things have happened
Many People Holding AI would have sold it by this time. They will have gone from holding AI Equities to not holding AI Equities and potentially wanting to return. They also may potentially be in the undecided group.
The only ones owning AI Equities at this point are the only ones left that will never sell. They are the high conviction AI holders. They may even start to buy more.
People shorting have maxed out their risk limits on the trade. No more shorts to make prices go down more, even in the face of negative sentiment.
When this happens, Shorts will close their position and there will be a sharp rise in AI Equities. Whether this is then picked up as a sign of recovery remains to be seen. The undecided and previous owners begin to crunch numbers, wrestling with both fear of losing money and FOMO. They will then come back based on price action so long as the news-flow remains ambiguous about where the AI narrative is going. Since we’re going into the summer, we may find a light news cycle.
Honestly, this is a fairly difficult concept to share. I’ll probably need to refine it in future articles. But the concept is fundamentally “watch out for news failure” I’ve discussed in previous articles. However, with so many discussions about the ramifications of META’s announcement, I felt it would be best to go into “watch out for news failure” in greater detail, highlighting the psychology of the different actors. I hope you can see how analyzing things this way (in conjunction with the fundamental assessments of the AI narrative), can help you make better decisions on your portfolio.
I hope you enjoyed today’s article. If you like my content, please consider liking and subscribing. Happy Trading!