I am growing tired of the endless bubble talk about all of the data center spending. It’s obvious from this quarter that the bubble talk has been proven wrong; try getting someone to say that, though. So, who do I think I am to go there? Just an observer. One who believes that this was the quarter where we realized that if you didn’t spend, you were already behind the 8-ball. This quarter, so far, we have seen the results of five large companies that are often discussed as creating the bubble: Club names Alphabet (parent company of Google), Amazon , Apple , Microsoft , and Meta Platforms (parent company of Facebook, Instagram, Threads, WhatsApp). Those are five of the “Magnificent Seven” stocks. (No. 6 is Nvidia , which we also own; it reports on May 20. Tesla is the seventh, but we don’t own it.) All so far have spent a huge amount to build out their sites. All have much more building to do. Let’s look at how much the companies estimated their capital expenditures to be this year and how their stocks did this week before and after earnings. Alphabet, data center spend, $180 billion to $190 billion, stock price: $349 to $385 for a 12% weekly gain. Amazon, data center spend, $200 billion, stock price: $260 to $268 for a 1.6% weekly gain. Apple, data center spend: $13 billion, stock price: $271 per share (April 24 close) to $280 (Friday close) for a 3.4% weekly gain. Microsoft, data center spend, $190 billion, stock $429 to $414 for a 2.4% weekly loss. Meta, data center spend, $125 bilion to $145 billion, stock price: $670 to $605 for a 9.8% weekly loss. Consider what they used that spending on: Alphabet: Google Cloud, tensor processing units (TPUs are custom chips co-designed by Club name Broadcom ), and graphics processing units (GPUs) Amazon: Amazon Web Services (AWS), Anthropic cloud capacity; custom Trainium, Graviton, and Inferentia semiconductors Apple: private cloud Microsoft: Azure companion, OpenAI compute needs Meta: Internal training, recommendation engines These are the broad-brush spend numbers, and we can learn a lot about the bang for the buck now that we are a quarter in full artificial intelligence motion. Here are some conclusions. Alphabet and Amazon have had some excellent post-earnings reactions. Why? I think that Alphabet’s stock is reacting to the incredible growth of Google Cloud and the perfect steerage of Thomas Kurian, who runs the cloud business. Google Cloud is the fastest-growing at 63% with an annual revenue run rate of over $80 billion. This quarter, it had $20 billion in revenue. Google also uses its spend to bolster the seamless transition from Google Search to Gemini. Amazon’s stock rallied because Amazon Web Services is now growing at 28% with an annualized revenue run rate of $150 billion and a quarter where it did $37.6 billion in revenue. That’s the fastest growth in 15 quarters, a huge deal when you consider the base. It wasn’t that long ago when AWS was growing in the low double digits. This growth shocked everyone and led, in part, to the great performance of the stock. Apple has the least amount of spend and it has a free-riding business off of Google. Apple may not have wanted that; in fact, it backed into it. But because Apple has a worldwide installed base of 2.5 billion devices, it earned the right to get Google’s Gemini for a low cost — some would say no cost when you add in how much Google pays for Google to be embedded in the Apple iPhone on the search side. Google can dominate because of Apple and not vice versa. Microsoft’s stock fell, I think, in part because while Azure is growing at 40% with an annualized revenue run rate of $90 billion to $95 billion and a quarter of $22 billion to $24 billion in sales, those numbers include compute demand from OpenAI. Again, we don’t know the breakdown, but the street isn’t giving it any credit for Azure’s acceleration from the 36% the Street expected. There’s also no certainty to their capex numbers; they could be higher. Microsoft is very vulnerable as a company because while it has a hybrid model of consumption and software as a service, the latter has gone from being a premium business to a business in search of the appropriate discount rate. Plus, unlike Google with Gemini, Copilot is not considered up to snuff, even with 20 million paid users. Unlike Amazon, with its web service and retail businesses as well as its advertising business, Azure hasn’t monetized its cloud business enough, and a lot of it is derived from OpenAI. Unlike Google, there’s, yes, Bing. Meta doesn’t have a cloud business at all, so it can’t monetize what it has as well as the other companies. Plus, it actually decided to increase data center spending by $10 billion. In other words, not only was its spending viewed as profligate to begin with, given that it doesn’t have a cloud, it actually is spending more with a very uncertain return, especially given that its Meta AI isn’t wowing anyone. So now let’s zoom out to see what the stocks are saying about the so-called spending bubble. Alphabet’s spend makes a ton of sense. Just as Google Search beat everyone, including Microsoft, it looks like it is happening again with Gemini. The spend is also worth it because of its giant YouTube business. It’s being appropriately rewarded. Amazon’s spend also seems totally justified and then some, given that AWS, its most lucrative division, is accelerating. The market’s wising up to the luck (and greatness) of Apple. Given its leverage on its newfound accelerated growth rate of 17% (from 9% previously), it deserves the high multiple it’s been given: twice its growth rate at $280. If Apple services keep growing at its 16% pace, the stock should go higher, given the incredible gross margin on that business of 77%. It prints money as it grows its business by offering multiple services for subscription that you can’t live without. The people who believe that all of this spend leads to a bubble do not understand the power of being number one. Google-Gemini is No. 1 in search of the publicly traded companies. Amazon is No. 1 in shopping, advertising, and enterprise cloud. Apple is No. 1 in phones. Meta is No. 1 in the still fast-growing social media business, but it is ad-supported — and if we go into a slowdown, it is vulnerable and may not make the numbers. It has the most cyclicality. Microsoft holds the number one spot in enterprise software. That was a terrific advantage for ages, but it has a huge “per seat” business, and that business could be disrupted by AI. It is considered very vulnerable. Plus, we may be at the high watermark for all of its enterprise software products, given what we know about OpenAI and Anthropic’s Claude. That makes us want to pay less for its stock as there might be fewer Microsoft users. So, now let’s simplify: Amazon and Alphabet are getting the most for their dollars and are unlikely to be disrupted by AI. Their spend is worth every penny. Microsoft may have to spend even more than it has to come up with better products that can keep its enterprise software dominance. The market is saying right now that it can’t. Meta’s in trouble and needs to make some sort of move to bring out shareholder value — or else, I am prepared to sell it for the Club portfolio. So far the “Trust in Mark Zuckerberg” view is not paying off. And, Apple’s got nothing but upside, no matter what happens. Or, to put it another way, he who spends the most and spends it well, wins. On Alphabet’s post-earnings call, CEO Sundar Pichai said, “Our cloud revenue would have been higher if we had more compute.” Everyone is compute-constrained and can’t get enough Nvidia. It fits right into what Nvidia CEO Jensen Huang has been saying for years: “Anthropic and OpenAI are highly compute-constrained. Doubling their compute capacity could increase revenue fourfold. Compute equals revenues.” With training, Jensen also correctly said, “The more you buy, the more you save.” With inference, it’s now, “the more you buy, the more you make,” Jensen concluded. Tell me how that spend isn’t incredibly important and justified, especially in a world where OpenAI could be worth a trillion, Anthropic maybe more than a trillion. Both are looking to become publicly traded companies, maybe as soon as late this year. Those two have an element of free riding, too, as their businesses are fast-growing and are therefore boosted by companies that want to sell something. The bubble talk only works if you don’t care about a company’s dominance or its stock price. I care about both. I want to make money. Not just talk about bubbles. As a lot of potential gains flow from these determinations, these companies are spending to have the best AI agents. These agents — software programs that can take instructions, think, plan, act, and learn — need massive amounts of central processing units (CPUs), which are made by the skyrockets that are Advanced Micro Devices , Intel , and our newest Club name Arm Holdings . AI agents are often used as virtual personal assistants, in call centers, and even for computer coding. Spending underlies the fiber buildout and connectivity companies like Lumentum , Coherent , and portfolio holding Corning. It helps the networking companies, Ciena , Cisco Systems , and Arista Networks (of which I like Arista best and wished I had bought when we exited Cisco). It bolsters the fancy GPUs from Nvidia, of which all but Apple still use; they just don’t talk about it much. For all of its bluster about Trainium and Graviton, Amazon is buying one million chips by the end of 2027. It means the world to chip partners Broadcom, which partners with Google, Meta, OpenAI, and Anthropic — and Marvell Technology , which collaborates with Amazon and Microsoft. These two companies— Broadcom and Marvell — help wean the big hyperscalers off a part of Nvidia and have hurt Nvidia’s stock the most, especially given that Anthropic runs mostly on Amazon’s chips; hence the big business there. It’s incredibly important to those involved in the building of the data centers, the improvement of the grid, and the cooling process: We’re talking about Quanta , Oracle , Vertiv , Nebius , and CoreWeave, as well as Club names GE Vernova and Eaton . It’s great business for back-up power generators like Cummins , Caterpillar , and Generac . And, of course, it is the lifeblood of memory companies Micron , SanDisk , Seagate , and Western Digital . The bubble callers would say that each one of these stocks is inflated. I come back and say that it is existential to those companies that don’t spend enough. The bubble callers say all of these stocks have run too much. I say that if they settle in here and work off their parabolic status, many of these could still be bought. The bubble callers think that this is 1999-2000 when the dotcom boom unravled. Let’s just refresh my standing: I was a hedge fund manager who created a company that at one point was worth more than a billion, and not that long after was worth $70 million (in enterprise value) — and, simultaneously, had his best years in the hedge fund business in 1999 and 2000, with the latter gaining 36% almost all on the short side. In other words, been there, done that, twice. Here are the key differences. In 2000, Google was private and could outspend everyone and won in search and video. That’s how it got to its exalted status. Amazon was spending a fortune to knock everyone out of online spending and then out of the web business; when it was done, its dominance in Prime. Apple hadn’t created the phone yet. Microsoft was prosecuted by the government for being so powerful. Meta hadn’t started yet. The spend then was worth it. Now on the vendor side then, not so good. This is where the bubble was. Instead of building out the power network, we built out the fiber network. Big mistake. We weren’t there yet with video. We only had dial-up. So, we didn’t need so much fiber. That was the bubble on the hardware side. On the software side were countless companies that competed with each other in all sorts of verticals that ultimately produced only one or two winners in each segment. Most were traffic winners. Those that could monetize on a subscription basis were winners. So were those who could attract enough eyeballs. What we didn’t realize was that the brick and mortar businesses after initially stumbling came roaring back to crush the online only companies with only a handful of Amazon-like companies surviving. Were there bubbles? Absolutely. How were they formed so easily? Because the syndicate desks didn’t see the rise of the individual trader, chiefly via E-trade and Ameritrade, and Schwab . It was so easy to put orders in, and there was so much money being made that the venture capitalists threw money at anything, and then it came public with very little vetting. It was such a ridiculous time that the accounts that did the most business and the friends and family accounts got stock and then could flip it when the retail buyers came in with market orders. The syndicate desks engineered the pop by only offering a sliver of stock. In six months’ time, if they were still in business, the companies did giant secondaries that included huge insider selling and small amounts of money to help the business. Then the businesses went under. More than 300 of them. Fast-forward to now, you see a lot of businesses coming public so far that look like the unprofitable losers of that dotcom bubble-bursting era? Many thought that CoreWeave looked like that, but CoreWeave is creating a business of tremendous value as a buyer of Nvidia chips, which more than hold their value. Could there be circular deals? Another canard. I was approached by a company that was later prosecuted for the lazy Susan deals it was doing all over the place. It wanted to give me a $10 million investment that I was immediately going to flip back to buy services the company was buying. I wish I were wearing a wire. I knew not to take that deal. We thought that Nvidia was doing circular deals all over the place. Now, what we see is just a series of incredible investments. The only miss was not giving Anthropic $10 billion when it first needed it. No Lazy Susans. Notice no one even talks about circular anymore? It is too absurd. So, let’s sum it all up. This quarter’s earnings reports amount to nothing short of a Judgment Day for AI. The companies that spent well are now the winners and are beginning to get the kinds of returns we once thought possible, with the exception of Apple, which got the return by free-riding off Google. If Apple hadn’t spent as much as it has to establish cellphone dominance, it wouldn’t be a winner. The companies that didn’t spend enough or didn’t spend wisely are now faced with a shrinking multiple and an uncertain future. Microsoft has a big installed base, it might be able to produce something important to justify the spend. Meta, too. But make no mistake about it, I am worried about both right now. It’s too bad that there were so many who didn’t believe or people who thought, “who cares, I am in an index fund.” Let me tell you something, as someone who can only own an S & P 500 index fund personally (because of my role as a financial journalist), I care. I don’t think you would have gotten to the end of this piece if you didn’t care, too. Correction: An earlier version of this story misstated the number of paid Copilot users. (Jim Cramer’s Charitable Trust is long AAPL, AMZN, GOOGL, MSFT, META, NVDA, AVGO, ETN, GEV. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust’s portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
Be the first to comment