Tech valuations slide back toward pre-AI norms as sector forward P/E falls to about 20x

A sharp reset in Big Tech valuations is pushing key metrics back toward levels seen before the AI-driven run-up. Apollo Global Management partner and chief economist Torsten Slok highlighted in the firm's Daily Spark (April 11) that the S&P 500 Information Technology sector's forward price-to-earnings multiple has fallen from roughly 40 at the height of the AI boom to around 20. The chart tracks the index's 10 largest constituents by market value: NVIDIA, Apple, Microsoft, Broadcom, Oracle, Micron, Palantir, AMD, Cisco and Applied Materials. The repricing is being blamed on a mix of forces rather than any single trigger. One is a rotation into energy and defensive areas following heightened Middle East tensions. With war in Iran cited as a catalyst, the energy sector rose more than 34% in the first quarter and ExxonMobil was up nearly 42% year to date, drawing capital away from technology. The broader S&P 500 also flashed a "death cross" late March, and by early April was hovering near 6,582, within about 100 points of the 6,300 level typically associated with a correction. A second pressure point is uncertainty over whether massive AI-related spending will deliver returns fast enough. FactSet data show the S&P 500's expected Q1 earnings growth at 12.6% with a forward P/E near 20.4x. Yet hyperscalers are committing extraordinary sums: Amazon plans to spend $200 billion by 2026, while Microsoft, Meta and others have mapped out investments in the hundreds of billions. A CEIBS research estimate cited in the brief suggests that to break even on roughly $400 billion of AI capex in 2025 alone, annual AI-linked revenue would need to reach $160 billion, versus an estimated $15 to $20 billion at the time. The third headwind is moderating profit momentum. Bloomberg Intelligence expects the seven major tech companies to post profit growth of about 18% in 2026, the weakest since 2022, narrowing the gap versus the 13% expected for the other 493 S&P 500 companies. UBS Global Wealth Management's head of U.S. equities David Lefkowitz said in January that earnings growth is spreading beyond tech, reducing the sector's dominance. At the single-stock level, dispersion is widening. Zacks data show NVIDIA's forward P/E down to about 21.4x, well below its 10-year median of 45.3x, even as annualized earnings growth of 39.1% is projected over the next three to five years. Microsoft is down about 23% year to date, and its market value has slipped below $3 trillion after topping $4 trillion in October last year. Apple has held up better within the Magnificent Seven, helped by comparatively lighter AI capex and aggressive buybacks: it repurchased $24.7 billion of shares in a single quarter, benefiting from investor preference for capital discipline as heavy spenders are penalized. Insider activity has added to unease. Motley Fool, citing SEC Form 4 filings, reported that insiders at NVIDIA, Apple, Alphabet, Microsoft and Amazon sold roughly $16.1 billion in net shares in the two years through April 2. While much of the selling is linked to tax-related compensation activity, the lack of insider buying alongside the net sales has unsettled parts of the market. The debate over an AI bubble is intensifying, with many drawing distinctions from the 2000 dot-com era. BlackRock said in a technology-sector report that the S&P 500 Information Technology Index traded at about 30x forward earnings in October 2025. That level is elevated but still well below the roughly 60x seen at the Nasdaq 100's dot-com peak, and BlackRock argues today's valuations are supported by real revenue, proven business models and accelerating AI adoption. Goldman Sachs has similarly noted that the long-term dividend growth implied by current prices is high, but below extremes seen during the internet bubble and the 1960s "Nifty Fifty" period. Valuation warnings remain prominent. The Globe and Mail reported that the S&P 500 began the year at its second-highest valuation level in 155 years based on the Shiller P/E ratio. Historically, when Shiller P/E exceeded 40—during the dot-com bubble and in January 2022—the index later fell 49% and 25%, respectively. Zacks offers a more tactical read: as prices fall while earnings expectations rise, multiples are compressing mechanically, improving the risk-reward setup for select names. NVIDIA is viewed as offering the strongest growth-to-valuation fit, while Microsoft is seen as a potential catch-up rebound candidate. For investors, the central issue is less whether AI is valuable and more whether today's capex can produce profits that justify valuations within a reasonable timeframe. If 2026 proves to be the cyclical peak in hyperscaler spending, the payoff period for AI infrastructure may stretch beyond market patience even as the technology continues to advance.