Artificial intelligence has helped push US stocks to new heights, but concerns about a possible market crash are growing. Rising valuations among major technology companies have raised questions about whether investor enthusiasm has gone too far. As fears of another dot-com-style bubble emerge, Warren Buffett’s long-term investment strategy offers an alternative to chasing expensive AI stocks.
AI valuations raise concerns
The S&P 500 has enjoyed several strong years, gaining 24% in 2023, 23% in 2024 and approximately 16% in 2025. Meanwhile, the technology-heavy Nasdaq Composite rose 43%, 29% and 20% during those respective years. Both indexes continued climbing in 2026, largely supported by enthusiasm surrounding artificial intelligence. However, much of this growth depends on a relatively small group of technology giants, leaving the broader market exposed to potential losses.
One warning sign is the cyclically adjusted price-to-earnings (CAPE) ratio, which currently stands near 41 compared with its historical average of around 17. Such elevated valuations have appeared during previous periods of market speculation, including the dot-com bubble. Investors now face a familiar question: can AI-related earnings justify these prices, or has excitement pushed valuations beyond reasonable levels?
Buffett’s long-term approach
Rather than attempting to predict market crashes, Buffett has consistently recommended low-cost index funds for ordinary investors. Back in his 2013 Berkshire Hathaway shareholder letter, the billionaire explained that 90% of his wife’s trust should be invested in a low-cost S&P 500 index fund, specifically recommending Vanguard.
Berkshire Hathaway later purchased 43,000 shares of the Vanguard S&P 500 ETF (VOO) in 2019, reflecting this investment philosophy. Historically, the S&P 500 generated annualised returns of approximately 10% between 1928 and 2025. Nevertheless, diversification cannot eliminate market risks. A sharp decline in major AI companies would still affect VOO because those businesses represent substantial portions of the index.
