The financial reports of the tech giants in the US stock market in the third quarter have been successively released, with the latest developments in AI prominently displayed in the reports.

For those concerned about AI trading, such as the situation of AI capital expenditures and the returns behind massive investments, what answers did the financial reports provide? What emerging trends in AI did the reports reveal?

Tech giants continue to increase AI expenditures without hesitation.

In the third quarter, major tech giants still adhere to the narrative of "increased spending" in AI. Microsoft, Google, and Meta Capital expenditures increased by 3% month-on-month and 48% year-on-year, pledging to continue increasing expenditures on servers and other AI-related costs in the coming quarters.

Specifically:

$Alphabet-C (GOOG.US)$ In the third quarter, capital expenditures were 13 billion US dollars (down 1% month-on-month, up 62% year-on-year), exceeding market consensus by 3%, mainly driven by general server replenishment and strong AI server expenditures. The company expects capital expenditures in the fourth quarter of 2024 to be similar to the third quarter, meaning overall spending in 2024 will reach approximately 51 billion US dollars, a year-on-year increase of about 60%. Management anticipates a substantial increase in capital expenditures in 2025, primarily driven by investments in AI infrastructure.

$Microsoft (MSFT.US)$ In the third quarter, capital expenditures reached $15 billion (an 8% increase from the previous quarter and a 50% increase year-on-year), exceeding market expectations by 3%, mainly due to the acceleration of the H200 AI server project and increased infrastructure investments. Due to investments in Azure and AI infrastructure, capital expenditures for the next quarter are expected to increase by 32% year-on-year. Capital expenditures for the 2024 fiscal year are expected to increase by over 55%, primarily driven by proactive AI investments and replenishment of general servers.

$Meta Platforms (META.US)$ In the third quarter, capital expenditures were $8.3 billion (a 1% increase from the previous quarter and a 26% increase year-on-year), lower than Bloomberg's expectation of $11 billion, due to variations in server deliveries. However, the company has narrowed its 2024 capital expenditure guidance range from the previous $37 billion to $40 billion (a 41% year-on-year increase at the midpoint) to $38 billion to $40 billion (a 43% year-on-year increase at the midpoint), primarily driven by investments in AI infrastructure. Meta's management has reiterated significant capital expenditures planned for next year in AI infrastructure.

$Amazon (AMZN.US)$In the third quarter, $22.6 billion was spent on real estate and equipment, an 81% increase from the same period last year. The company's capital expenditures have reached $51.9 billion since the beginning of the year. The company anticipates capital expenditures for 2024 to be $75 billion, approximately 55% higher than 2023's $48.4 billion. Capital expenditures are being driven by the AWS cloud division, and capital expenditures for 2025 may be even higher.

In response, JPMorgan analysis believes that:

From the financial reports of Google/Microsoft/Meta, optimistic capital expenditure comments, strong core businesses, and positive prospects in AI business can be seen. This may enhance investors' confidence in the sustainability of AI expenditure and alleviate concerns in the market about the impact of AI server consumption. It is expected that optimistic spending guidance from Google/Microsoft/Meta will drive short-term sentiment in downstream AI servers.

The return behind the massive investment is "mixed".

For the AI capital return that the market has been focusing on, tech giants have seen some positive growth signs in conversational AI services, but this growth seems to have not reached the expected level, and may be offset by spending in other areas, limiting the overall growth of tech companies. Shareholders will have to wait longer to know the specific return answer.

Microsoft reported a slight decline in growth in the Azure cloud computing department in the third quarter, with a larger decline expected in the fourth quarter. However, Microsoft stated that the slowdown in Azure's growth is not related to reduced customer demand, especially in this quarter, but more due to the company continuing to face capacity constraints. Analysts predict that starting in the second half of 2025, with more AI servers coming online, Azure's growth will accelerate, indicating that Microsoft's AI server supply chain revenue will accelerate over the next few quarters.

On the positive side, the demand for AI products is increasing, and Microsoft expects revenue from all of its AI products (such as OpenAI models sold through Azure, GitHub Copilot, and Office Copilot subscriptions) to exceed $2.5 billion per quarter in the next two months, with Microsoft's total revenue in the third quarter reaching $66 billion.

In addition, Google's report states that cloud computing department growth is accelerating, with Google's cloud revenue in the third quarter growing by 35% year-on-year, an acceleration from the 29% growth in the previous quarter. This is mainly due to strong core/AI GCP demand and Google Workspace revenue expansion, as well as improving cross-industry advertising demand indicating a strong momentum in general server spending.

However, this is partly attributed to factors unrelated to AI. Earlier this year, Google Cloud began executing contracts with some of its largest customers, who committed to spending a certain amount of revenue annually to lease servers. People involved in these transactions stated that in some cases, customers pay a lump sum for the unused cloud services they had previously committed to purchase, signifying a departure from past practices where Google Cloud allowed customers to carry over their expenditure commitments to the next year.

"Hiring is not as good as buying cards", training shifts to reasoning, and Microsoft marginalizes OpenAI.

This financial report also reveals three major trends: one is that "hiring is not as good as buying cards", the second is the shift in training to reasoning, and the third is Microsoft "diluting" the importance of OpenAI.

Firstly, despite microsoft's significant increase in investment in artificial intelligence, on the other hand, it is also trying to control expenses as much as possible, one way being through layoffs and improving employee efficiency. Specifically, since 2022, amazon has laid off tens of thousands of employees. Expense control has slowed the total operating cost growth rate from 17.6% in Q3 2022 to 7.3% in Q3 2024.

Secondly, the importance of inference goes beyond training, meta platforms have officially decided to prioritize inference work over AI training. Microsoft expects the annual revenue of its AI business to reach $10 billion in the next quarter. CEO Nadella of microsoft stated that this growth is driven by inference rather than training. Nadella also mentioned that microsoft does not sell GPUs to others for training, as there is a great demand for inference to support Copilots and other AI services.

Analysis indicates that the key to microsoft's strategic modification is its decision to reject customers who are eager to rent GPUs for training new AI models, instead microsoft's goal is to utilize these resources for inference. As more and more enterprises integrate AI components into their operations, the demand in this area is thriving. This shift demonstrates a keen understanding of market dynamics. Compared to model training, inference can provide more direct and stable income.

Lastly, it is worth noting that microsoft is reducing the importance of OpenAI. Some analysis suggests that microsoft has announced the introduction of Claude 3.5 Sonnet and Gemini 1.5 pro, indicating that OpenAI is no longer the only choice. This indicates that microsoft aims to architect more orchestration layers on top of the models to capture greater value from the middle orchestration layer, similar to the cloud computing era, such as the recently released Agent creation tool. The value of the bottom model layer, which is OpenAI, will be downgraded.

Editor/Rocky