Big Tech’s massive expenses in artificial intelligence should continue without control in 2025 after Amazon exceeded its competitors with a planned investment of $ 100 billion in infrastructure this year.
The expenses of the four main American technological companies had already jumped 63% at the historic levels last year. Now, leaders promising to accelerate their investments in AI, rejecting concerns about the vast sums betting on emerging technology.
Microsoft, Alphabet, Amazon and Meta declared combined capital expenses of 246 billion dollars in 2024, against $ 151 billion in 2023. They predict that spending could exceed $ 320 billion this year because they compete to build centers of data and filling them with clusters of specialized fleas to stay at the forefront of research on large language models of AI.
The extent of their spending ambitions – announced in parallel with their fourth quarter income – surprised the market and exacerbated a sale caused by the release of an innovative and cheap AI model from the Chinese start -up Deepseek at the end of January .
Microsoft and Google Parent Alphabet have each seen $ 200 billion suffered from their market value after reporting lower than expected growth in their cloud computing divisions in parallel with a sharp increase in capital expenditure. The 8% drop in Google Wednesday was its fifth day of negotiation in the last decade.
“The frantic enthusiasm in the whole” Magnificeent Seven “was replaced by pockets of skepticism and created situations” show me “,” said Jim Tierney, chief of the American growth fund concentrated in Alliancebenstein. “The concerns that I have not had since the summer are amplified today.”
In the midst of the media threw on the potential transformer of the AI, the shareholders fear that the doubling of expenses without a proportional increase in income can eat in a capital which would have otherwise returned in the form of redemptions and dividends, while falling out companies not.
Google was opaque regarding the use and income of its Chatbot Gemini, while businesses are wary to adopt Microsoft “co-pilot agents” to improve labor productivity.
“If or when we see the acceleration of cloud growth at Google or [Microsoft’s] Azure, or see co -pilot absorption improve, investors will be more comfortable with spending at Alphabet or Microsoft, “said Tierney. “Cheaper and more merchant AI models will likely amplify the concerns of investors in the meantime.”
Deepseek’s R1 model was emblematic of such fears. The pretension of AI Lab Chinese to have built a reasoning model with capabilities similar to the products of Google and Openai to a fraction of the price – and without access to the most advanced graphic processing units of Nvidia – has plunged the stock of the flea manufacturer, efforcing $ 600 billion in a day, from which he only recovered.
The great chiefs of technology kept their nerve. Tuesday, Sundar Pichai of Google declared in defense of its plan to spend $ 75 billion in 2025 – up 42% compared to $ 53 billion last year – that the opportunity of the AI was “also Large that, and that is why you see us investing to invest to meet this moment ”. Deepseek would add to demand by showing how new techniques could make it cheaper and stimulate new research lines, he said.
Satya Nadella from Microsoft said two weeks ago in Davos two weeks ago: “I will spend $ 80 billion to build Azure, customers can count on Microsoft.” He reiterated his belief in the madness of the slowdown and failing to capitalize on his early support from the start-up Openai.
And Thursday, Amazon CEO Andy Jassy exceeded Google and Microsoft by providing more than $ 100 billion in capital spending this year, compared to $ 77 billion in 2024 and more than double dollars of the previous year. The vast majority will go to data centers and servers for Amazon’s web services, and Jassy said it simply responded to “important demand signals”. The action dropped up to 7% in exchanges after opening hours.
“Growth cooks a bit, but investment appetite has not been reduced,” said Jeff Pearson, vice-president of the Cloud strategy at Consultancy Presidio. “They plow even if the return on investment seems distant.”
Meta received a more positive reception for its profits, its actions increasing while chief Mark Zuckerberg has committed to spending “hundreds of billions” more on AI, in addition to the $ 40 billion invested in 2024.
“Investors have adopted Meta, even if their capex increases, because there is an improvement in real -time return on investment in customer spending that is measurable,” said Tierny, referring to the use of Meta de l ‘IA to improve the targeting of ads on Facebook and Instagram.
Meta’s success in the watch of tangible yields of investment in AI contrasts with Google, which faces new competitors and the difficult task of integrating AI into research without cannibalizing its basic advertising activity.
The research giant has introduced brief responses, or “IA seen”, at the top of research results, but they move its lists of links, the first of which are often sponsored by lucrative.
Nevertheless, “if he is supposed to be cracks in the Google research empire, he certainly does not present himself,” said Bernstein analyst Mark Shmulik, indicating 13% growth in advertising revenues $ 54 billion in the last three months of 2024 alone. “Google has not missed research expectations even once since the launch of Chatgpt a nine quarters.”
Expenditure among the “Magnificent Seven” – which also include Apple, Nvidia and Tesla – Dwarfs The rest of the American reference S&P 500. Their capital expenses increased by 40% in 2024 against 3.5% among the 493 companies remaining, According to General Society. The benefits among the elite group have skyrocketed a third in the same period, against 5% among the others.
The spending wave is not limited to listed companies on the stock market, and neither the deep research nor the fears of a bubble of AI slowed down the capital flow in the Silicon Valley start-ups.
Sam Altman of Openai has trained a partnership with Softbank and Oracle to invest $ 100 billion in the US infrastructure linked to AI, potentially increasing half a bill over time. The Japanese investor is in talks to invest $ 25 billion in the start-up in an assessment of $ 260 billion.
“Could there be a winter of AI at some point?” Of course, ”said Rishi Jaluria, analyst at RBC Capital Markets. “But if you are able to be a leader, you cannot remove the foot from the gas.”
