Goldman Sachs re-discusses the "AI Trading Four Stages": The second stage "AI Infrastructure" is nearing maturity, don't overlook the application in the third stage

Over the past six months, the fluctuation of AI-related stocks has intensified. After a recent rebound, Goldman Sachs pointed out in its latest research report that the "second stage" of AI trading will tend to mature, and investors need to be cautious about investing in the "third stage".

Six months ago, we quoted Goldman Sachs' "Four-Stage Theory" on AI investment in an article. Among them, as the most clear beneficiary of AI, NVIDIA belongs to the "first stage", the "second stage" includes companies focusing on AI infrastructure, the "third stage" focuses on companies with the potential to monetize AI through incremental revenue generation, and the "fourth stage" mainly involves companies with the greatest profit potential in the widespread adoption and productivity improvement of AI.

Goldman Sachs believes that the current investment focus is on the second stage of infrastructure, with typical stocks in this category having risen by 27% so far this year, valuations exceeding the 10-year average by 0.4 standard deviations, gradually entering an overvalued stage. Goldman Sachs warns that future growth of stocks in this stage may rely more on actual profit performance rather than further valuation expansion:

We expect the second stage of trading to mature, with stock price increases being driven by earnings growth rather than valuation expansion.

Starting valuations are above average, NVIDIA's sales exceeding expectations, capital expenditures of large-scale cloud computing service providers exceeding expectations, and the growth rate of enterprise mentions of AI have all slowed down.

The upcoming third-quarter earnings season will be an important test.

For AI application stocks in the third stage with the potential to generate incremental revenue, Goldman Sachs advises caution. However, Goldman Sachs believes that in this stage, "platform" stocks show strong investment appeal:

Although valuations are below average, the timing of AI application landing and monetization remains uncertain.

Investors confident in the subsequent stages of AI development can choose some "platform" stocks.

Second Stage: Growth Challenges in Overvalued Conditions

Currently, the investment focus is mainly on stocks in the second stage of infrastructure. These stocks have risen by 27% so far this year, with valuations exceeding the 10-year average by 0.4 standard deviations.

Among them, the massive AI investments by tech giants are a key factor supporting these stocks. Tech giants such as Amazon, Google, Meta, Microsoft, and Oracle have announced large-scale AI-related investments, with capital expenditures expected to reach $215 billion and $250 billion in 2024 and 2025, respectively, with a significant portion related to AI.

However, overvaluation also brings growth challenges. Goldman Sachs analysts point out that the future growth of these stocks may rely more on actual profit performance rather than further valuation expansion:

We believe that there is still room for growth in second-stage stocks, but future increases will be more driven by earnings growth rather than valuation expansion.

The current stock prices of these companies have exceeded expected earnings per share, reflecting investors' optimistic expectations for AI potential. Some second-stage stocks have recently risen due to positive AI-related news (such as increased shipments by AMD, and Constellation Energy signing a nuclear energy agreement with Microsoft) However, due to the high initial valuation, future returns may be limited, although the predictive ability of valuation for the short-term performance of large-cap stocks is limited.

It is worth mentioning that the company also mentioned in the report that the magnitude of AI spending exceeding expectations is narrowing, and the frequency of mentioning AI by enterprises is stabilizing, which may imply that the future increase in stock prices in the second phase may be more moderate:

AI spending exceeding expectations is narrowing, implying that the future increase in stock prices in the second phase may be more moderate. Although the current valuation is slightly high, if the AI investment of tech giants exceeds expectations, it may still drive stock prices up. In early 2023, Nvidia's chip demand greatly exceeded expectations, and in the first half of 2024, cloud giants' capital expenditure expectations continued to rise. However, these exceeding expectations are narrowing recently. The upcoming third-quarter report will be a crucial test.

Similarly, although the frequency of mentioning AI by enterprises and the willingness of management to spend on IT are still higher than historical levels, they have stabilized. The proportion of companies mentioning AI in the second quarter of 2024 has slightly decreased compared to the first quarter. Our IT spending survey in September also shows a slight decrease in spending willingness compared to January.

Phase Three: Cautiously Select Application Stocks

Currently, AI application stocks in the third phase are trading sideways with valuations below the average. However, Goldman Sachs believes that the timing for AI application development and monetization remains unclear, making it difficult to fully deploy in the short term:

Based on our conversations with investors and recent stock price performance, most third-phase stocks with potential AI-enabled revenue have not met expectations in terms of application development and monetization. We expect this uncertainty to persist in the short term.

According to Goldman Sachs' IT spending survey, management expects only 3% of next year's IT budget to be allocated to generative AI. By 2025, there may be an improvement in willingness to spend on AI products, as by then management will have a clearer understanding of the direction of the Federal Reserve policy and election results.

However, the company believes that in this phase, "platform" stocks show relatively strong investment attractiveness:

"Platform" stocks may represent an attractive subset of stocks in the third phase.

Our analysts believe that "platform" stocks, including databases and development tools, will be the primary beneficiaries of the next wave of generative AI investments.

These platforms can both optimize AI infrastructure and provide basic modules for building next-generation applications.

Phase Four: Time Needed

For fourth-phase stocks, companies expected to achieve the greatest productivity gains from widespread AI applications, Goldman Sachs analysts believe that their performance will take time.

Data shows that only 6% of enterprises currently use generative AI in production, a proportion far below market expectations. Goldman Sachs analysts point out that only after substantial progress in the promotion of AI applications in the third phase can investors build confidence in these types of stocks:

We believe that the launch of applications in the third phase of stocks is a necessary condition for investors to build confidence in companies with fourth-phase stocks (those stocks with the greatest potential profit growth due to AI-related productivity improvements)