A war we can't afford to lose: the AI arms race of the tech giants

1

The second quarter earnings reports of the seven major US stocks, in addition to Nvidia, have all announced, from the perspective of profit growth alone, the highest Tesla 28%, the lowest Apple 10%, although in the case of the United States economy facing a recession, it is already very good, but compared with the stock price increase and valuation, it is not much good.

The real focus is hidden in the capital expenditures of these giants, with the "three clouds" (Google, Microsoft and Amazon) and Meta investing the most aggressively:

Q2 Capital Expenditure:

Google: 13 billion, up 91% year-on-year and 10% month-on-month

Meta: 8.5 billion, up 33.30% year-on-year and 26% month-on-month

Microsoft: 19 billion, up 78% year-on-year and 35.70% month-on-month

Amazon: 17.6 billion, up 54.39% year-on-year and 18.12% month-on-month

Much of the skyrocketing capital expenditure has been invested in AI infrastructure – from data center construction and expansion, to GPU procurement and custom chip work, and more.

Microsoft, the fiercest of them, as an example, had a profit of only 22 billion in the second quarter and a cloud computing revenue of only 28.5 billion, so it dared to smash most of it into "AI infrastructure".

Coupled with Apple's 4.9 billion and Tesla's 2.2 billion, plus NVIDIA's own R&D costs and capital expenditures, the capital expenditure of the Big 7 in the second quarter is between 650~70 billion, assuming half of it is used for the AI arms race, and the Big Seven alone will burn more than 30 billion in one quarter.

This is still in the second quarter when there are almost no big AI explosive applications and large model performance upgrades, otherwise the investment would have exploded long ago.

Many people exclaim that the investment in AI is overheated, and it is likely that nothing will come out in the end, which seems rational, but ignores that this is not an ordinary competition for technology products, to some extent, it is an arms race between AI giants that cannot be lost.

2

The valuations of big tech stocks are not as frothy as they are too high, but that's only before the investments don't generate new business and profits. According to the nonlinear characteristics of artificial intelligence input and output, maybe it will be the "last shovel", and then tens of billions of dollars will be invested, and the epoch-making products will be "burned out with money".

The main reason why it is said to be an AI arms race is that large companies are worried about missing out on this industrial revolution, and for tech giants, this is indeed a battle that cannot be lost. It's like the meaning of Google and Meta's shareholder meetings - "it's better to overinvest than lag behind".

There are three reasons why large companies are willing to take this "big gamble" without knowing the future:

First, I'm not afraid of losing gambling, but I'm afraid that I won't be able to play the game

Although the capital expenditure of the Big Seven is crazy, it is within the range of cash flow of its own business, and it does not use too much financing, assuming that everyone burns money together, and in the end nothing is burned, then the result is nothing more than a loss for everyone, nothing more than a bad performance in the next few quarters, everyone's stock price dives together, and what new business can be borrowed to run out in a few years.

But if you fall behind in the investment stage now, the stock price will now lose its imagination and collapse, and both investors and customers will lose confidence in you, which is equivalent to losing now. If you really miss it, it is not a matter of losing market share, but that a giant may not exist in the future.

Therefore, for big manufacturers, this game is a "coward game", which can be lost, but cannot be exited early.

Second, large factories have a higher chance of winning than small companies

This is not to say that epoch-making AI products are more likely to be developed by large companies, but in terms of the certainty of input and output, large manufacturers are much higher than small companies or even luxury startups.

The main reason is that the direction of AI breakthroughs is uncertain, and small companies are generally a single product, and the revenue that can be generated in the end may not necessarily cover the investment. However, for large manufacturers, its product line is rich and there are many application scenarios, and the output effect that the current investment may bring in the future will be more obvious, especially for these cloud computing vendors, as long as there is a breakthrough in AI, whether they have a business or not, they will benefit.

And, in any case, at least the effectiveness of AI in reducing labor costs is already very clear.

Third, Nvidia has contributed to the fire

The crazy investment of the Big Six is, to a certain extent, related to Lao Huang's "selling anxiety" on different occasions, Lao Huang said a bunch of AI future scenarios, which is actually a sentence that training institutions like to say: "You come, we train your children, if you don't come, we cultivate your children's competitors".

Therefore, for large manufacturers, they know that Lao Huang is "selling anxiety", but as long as the opponent does not stop, there is no reason for them to stop, and the follow-up investment in the entire development and infrastructure will still increase steadily from the previous month.

Of course, I don't think they will be too aggressive either, and there is no need to worry about getting into a "prisoner's dilemma" at the moment, leading to financial collapse.

Taking Meta's earnings report as an example, the capital expenditure in Q2 was actually lower than market expectations, but after it came out, the stock price still rose sharply. Zuckerberg had suffered losses before, and in the first quarter of late April, in order to show its investment in AI, Meta raised its capital spending this year, but Wall Street was worried that the huge investment would not be absorbed by revenue and profit growth, and the stock price fell 15% after the earnings report.

So this time, Meta raised the outlook for the lower end of the range for full-year capex by $2 billion, and the upper end of the range has not changed, which means, I will throw money, but not brainlessly.

This is the mentality of the current market, which not only wants AI to run, but also wants AI not to eat grass, and the technology giants know the mystery, basically "not to be the head, not to lag behind", the result is that everyone's capital expenditure grows together, so that this round of technology stock bull market has been in small steps, always overvalued and not too much bubble.

3

Back to the choice of investment targets, the most flat investments are, of course, the Nasdaq ETF and the S&P 500 ETF, but these are a bit problematic.

And I suggest that even if it is a global allocation, there is still a part of the A-share position, and the most relevant thing in the A-share sector is the mapping of the AI hardware industry chain, whether it is 2B servers, computing equipment, or 2C AI mobile phones, AIPC, as long as it is the electronic communication industry, it is inseparable from China's manufacturing industry.

Therefore, if you believe that the giants' "AI arms race" will continue to fight until it comes to fruition - of course, the results represent greater investment, then, in addition to U.S. stocks, you can also pay attention to the mapping sector of A-shares.

As I said before, I am most optimistic about the application explosion of GPT-4o in the direction of Apple mobile phones at the end of the year, so the most priority direction is the consumer electronics ETF (symbol: 159732).

The valuation level of consumer electronics is directly related to technological change, this direction from the wireless Bluetooth headset, there has been no explosive application of five years to bring big opportunities, if the AI can break out on the application side, the requirements for mobile phone configuration increase, this will bring a big opportunity for the entire sector.

The second choice is the chip ETF (code 159995), computing power chips are the biggest bottleneck in the development of domestic AI, if there is an explosive application in the United States, it will stimulate the accelerated development of the country, and the semiconductor industry itself is going through a new round of upward cycle, which belongs to the dual logic of "industrial cycle + technology catalysis".