Since 2025, AI trading has experienced significant volatility, with the rise of DeepSeek impacting the U.S. stock market. This has led investors to question whether the AI trading boom is over. However, Goldman Sachs analysts, including Ryan Hammond and David J. Kostin, suggest that despite current challenges, the ongoing advancement in AI technology and profit growth will eventually draw investors back.
Goldman Sachs categorizes AI stocks into four stages. The "first stage" includes AI chip stocks like Nvidia, the "second stage" consists of AI infrastructure-related stocks, the "third stage" involves companies generating revenue through AI technology, and the "fourth stage" includes companies benefiting from AI-driven productivity enhancements.
Recently, AI trading has been hit by downgraded economic growth expectations and position adjustments, leading to concerns about corporate AI spending capabilities. Since February 19, Nvidia has underperformed the S&P 500 equal-weight index by 16 percentage points. AI infrastructure stocks and AI revenue-generating stocks lagged by 9 and 7 percentage points, respectively, while AI productivity stocks fell the least, trailing by 3 percentage points.
Despite these challenges, Goldman Sachs remains optimistic about AI's long-term prospects. They believe that sustained technological progress and profit growth will eventually refocus investor attention on AI stocks. In the short term, market conditions require position clearing or improved economic data to change the situation. Goldman Sachs' sentiment indicator recently fell to -0.4, indicating that positions are not yet low enough to support a tactical rally.
Goldman Sachs finds third-stage AI stocks more attractive than second-stage stocks. As large companies' AI capital expenditure growth slows and AI costs decline, investor focus is expected to shift from infrastructure to technology application and revenue generation. Although second-stage AI stocks have already experienced a sell-off, their relative valuation remains slightly above historical averages, whereas third-stage AI stocks are relatively undervalued.
Notably, third-stage AI stocks have seen positive sales revisions, with their 2026 sales forecasts revised upward by 0.3%. In contrast, second and fourth-stage AI stocks have faced negative revisions, with their 2026 sales forecasts adjusted downward by 0.3%. Among the third-stage AI stocks identified by Goldman Sachs, companies like Palantir (PLTR, Financial), Cloudflare (NET), SentinelOne (S), and GitLab (GTLB) are expected to achieve the fastest sales growth over the next two years.
