news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

New Intelligence Report

Editor: Qiao Yang

【New Wisdom Introduction】The GenAI craze is evolving into a veritable "arms race". The huge capital investment and long payback cycle are undoubtedly a double test for technology giants and investors.

Microsoft's recently released financial report once again revealed the truth about the cost of GenAI.

The report showed that Microsoft spent a full $19 billion on cash capital expenditures and equipment purchases this quarter, a year-on-year increase of 78%, equivalent to the total expenditure for an entire year five years ago.

Not surprisingly, almost all of that $19 billion was related to cloud and AI, with about half going toward building and leasing data centers.

For the entire fiscal year 2024, Microsoft's total capital expenditures will account for approximately 23% of full-year revenue, compared with an average of only 14% over the past five years.

The funds spent in the field of GenAI have also rapidly increased overall spending. In fiscal year 2024, Microsoft spent a total of $55.7 billion, a 75% increase from the previous year.

But Microsoft seems confident in the path it has chosen, with CEO Satya Nadella saying on a conference call on Tuesday that they have captured "demand signals" to justify the investments.

And if those signals change unexpectedly, they can always adjust their plans, cut costs, and equip their data centers with ordinary servers instead of luxurious Nvidia chips.

According to Chief Financial Officer Amy Hood, this number will continue to increase in fiscal 2025, and these investments are necessary to support AI services.

Not only is spending surging, but Microsoft's Azure cloud service is also showing signs of weakness.

Just as Nvidia has made a fortune by "selling shovels", Azure has been Microsoft's main growth engine in recent years, but its revenue this quarter failed to meet analyst expectations.

Azure's revenue grew 29% in the fourth quarter, down from 31% in the previous quarter and below the expected 30.6%, with AI-related services contributing 8 percentage points.

For the full fiscal 2024, Azure's total revenue is $36.8 billion, about 2% below expectations.

Also in the data center sector, Nvidia is doing quite well. Last quarter, their related revenue reached US$22.6 billion, a 23% increase from the previous quarter and a 427% increase from the previous year.

Compared with Azure, the performance of Microsoft's other business units exceeded analysts' expectations, with annual operating revenue reaching a record $109 billion and a profit margin of 44.6%, a level that has not been achieved in the past two decades.

According to Microsoft's internal analysis, the recent cloud revenue shortfall was largely driven by weakness in several European regions, and Azure growth is expected to pick up in December, with plans to achieve double-digit operating revenue growth again next year.

Apart from Azure's understandable failure to meet expectations, Microsoft's performance throughout the fiscal year can be described as remarkable, with its merits outweighing its flaws.

But Wall Street investors don't have that patience. Following the release of the earnings report, Microsoft's stock price fell 7.8%.

Technology giants all saw their share prices fall to varying degrees at the end of July.

“Wall Street doesn’t have a lot of patience,” said Daniel Morgan, senior investment manager at Synovus Trust. “They see you spend billions of dollars, so they want to see corresponding revenue increases.”

In stark contrast, Chief Financial Officer Amy Hood believes that the assets Microsoft is currently investing in will be profitable over the next 15 years or even longer.

How expensive is AI?

Microsoft is not alone. Other technology giants are also burning money and have anticipated a long period of return on investment.

Google, once famous for its employee benefits, has also begun to drastically cut spending in the past one or two years, but has increased its investment in GenAI.

Yet the results were "not exactly exciting," in the words of Jefferies analyst Brent Thill. Google's overall revenue for fiscal 2024 was just 0.6% above the consensus forecast, the worst performance in at least the past five years.

Alphabet said on Tuesday that investment will continue in the second half of the year, with capital expenditures likely to reach or even exceed $12 billion per quarter, and total annual spending expected to exceed $49 billion, 84% higher than the average of the past five years.

Interestingly, Google CEO Sundar Pichai seems to see eye to eye with Nadella.

When asked about AI investments during the earnings call, he said: "We are in the early stages of a very transformative field. For us, the risk of underinvesting is much greater than the risk of overinvesting."

Meta is also burning money, but its profitability is clearly far ahead of Microsoft and Google.

In the last quarter, Meta's revenue was slightly over $39 billion, a year-on-year increase of 22%, of which net profit was approximately $13.5 billion, a year-on-year increase of 73%.

Zuckerberg said at the earnings conference that the number of users of Meta AI assistant is expected to surpass all similar products by the end of the year. Moreover, GenAI's real revenue will come from commercial use cases, such as making advertisements from scratch and allowing companies to create customized AI agents in WhatsApp.

At the same time, he also clearly warned investors: Although there are many ways to create businesses through GenAI, it is costly and takes a long time to generate profits.

Regarding the necessity of huge investments, his statement was almost exactly the same as what Google CEO Pichai said: It is difficult to predict how this will affect future generations, but at this point, I would rather take the risk of building excess production capacity than fall behind.

These Silicon Valley technology giants have strong financial support. Even if they cannot achieve strong revenue growth like Meta, at least they will not fall into a situation of cash flow shortage.

The situation is different for startups like OpenAI and Anthropic.

The Information did some calculations for OpenAI last week: If the current trend continues, they

Without additional capital injection, the company may face the risk of cash flow drying up as early as the end of the year.

Wall Street's fragile nerves

In stark contrast to the optimism and determination of technology companies, Wall Street's reaction fully reflects the sharpness and caution of capital.

More and more analysts and investors are beginning to suspect that the huge amounts of money being invested in AI will eventually turn into a financial bubble.

In the past few weeks, institutions such as Goldman Sachs, Barclays and Sequoia Capital have released reports stating that GenAI is currently unable to achieve profitability that matches its investment.

Jim Covello, a senior analyst at Goldman Sachs who has studied technology companies for 30 years, said in a report that "overbuilding things that have no use or that we are not ready to use usually does not end well."

Just over a year ago, Goldman Sachs also published a report stating that AI could automate 300 million jobs and increase global economic output by 7% in the next 10 years.

Vineet Jain, CEO of AI and data management company Egnyte, said the cost of developing and running AI programs will fall as other companies begin to compete with Nvidia and the technology becomes more efficient.

At present, the cost of providing AI products is too expensive. For example, one of the important reasons why OpenAI has such a large funding gap is that it has invested too much in inference computing power.

As costs fall and demand continues to rise, this current situation that makes investors pessimistic is likely to change.

During this difficult transition period, large companies such as Google and Microsoft can continue to invest with their strong financial resources, but small start-ups that are extremely dependent on venture capital may find it difficult to survive.

This seems to be a wave of emotional backlash and rational reflection on the entire AI craze in 2023. Jain pointed out this trend with a vivid metaphor:

“It’s like a soufflé that keeps popping up. It has to come down a little bit.”

References:

https://futurism.com/the-byte/microsoft-losing-money-ai

https://www.theverge.com/2024/7/31/24210786/meta-earnings-q2-2024-ai-llama-zuckerberg

https://www.bloomberg.com/news/articles/2024-07-30/microsoft-reports-slower-azure-cloud-growth-shares-drop?srnd=phx-ai

https://www.reuters.com/technology/microsoft-beats-quarterly-revenue-estimates-2024-07-30/

https://www.washingtonpost.com/technology/2024/07/24/ai-bubble-big-tech-stocks-goldman-sachs/