Are Earnings in an AI-Fueled Bubble?

Investors are asking whether an earnings bubble is forming after artificial intelligence demand pushed corporate profit forecasts sharply higher and helped lift US stock valuations to roughly 20 times forward earnings.

Wall Street analysts forecast about 25% earnings growth for 2026 and roughly 18% for 2027. Upward revisions are concentrated in technology, communication services and parts of the energy sector. Technology earnings forecasts have risen by more than 30% this year, and communication services estimates have climbed by over 20%. Analysts report that upgrades have been spreading beyond a handful of mega-cap firms into other industries.

Much of the upgrade momentum traces to semiconductors. Chipmakers have reported stronger sales and higher margins tied to AI demand. Major manufacturers and foundries have announced increased capital spending: Taiwan Semiconductor Manufacturing Co. has signalled roughly 40% higher capital expenditure, Samsung has outlined about $73 billion in investment for factory upgrades and research, and memory makers including SK Hynix and Micron are expanding capacity. Most new production is not expected to be online until 2027 or 2028, which market participants say should keep supply tight for the next 12 to 18 months.

Revenue projections for the memory sector illustrate the scale of the cycle: roughly $200 billion in 2025, about $600 billion in 2026 and nearly $800 billion in 2027. Some analysts caution that rapid increases in revenue and capacity plans may rely on optimistic assumptions about sustained demand and disciplined expansion.

Market concentration has increased. Data compiled by large banks shows an “AI Big 10″—including Nvidia, Microsoft, Alphabet, Amazon, Meta Platforms, Apple, Tesla, Broadcom, Micron and AMD-accounts for about 41% of the S&P 500. The Nasdaq Composite rose 21.4% in the second quarter of 2026, its strongest quarterly gain since the post-pandemic rebound, driven largely by AI infrastructure investment, semiconductor strength and major listings.

Valuation indicators give mixed signals. The S&P 500 trades at about 20 times forward earnings, below peak levels seen during the 2020 recovery and well below dot-com extremes. The cyclically adjusted price-to-earnings ratio for the S&P 500 has climbed above 40. Some analysts point to higher current profits at leading AI firms while also noting that investors may be paying premium prices for earnings that are elevated relative to history.

Several risks that could test current forecasts are being monitored by market participants. Renewed inflation could prompt the Federal Reserve to keep interest rates higher for longer, weighing on consumer demand. Rising oil prices would raise costs across the economy. A slowdown in corporate earnings growth, if it occurs while investors reduce the valuation multiples they are willing to pay, could produce simultaneous earnings downgrades and multiple compression.

Industry capital spending on AI infrastructure is shifting how large technology companies use cash. Firms are directing more capital into data centers and related projects rather than returning it through buybacks. Analysts estimate data center investment now represents more than 2% of US gross domestic product through new capital expenditure, creating demand for construction, electrical work, logistics and materials and supporting earnings in sectors beyond pure technology.

Arun Sai, senior multi-asset strategist at Pictet Asset Management, described the situation as “the strongest earnings upgrade cycle since the commodity supercycle.” Ben Inker, co-head of asset allocation at GMO, warned that forecasts for the next year’s profits rose almost 20% in six months, a pace he said has not been seen since 2021, and cautioned that those gains may not materialize. Capital Economics flagged the possibility that AI-related markets could reach levels where earnings expectations and capital expenditure assumptions become difficult to sustain. Sarah Ketterer, chief executive of Causeway Capital Management, observed that “low valuation multiples may not necessarily indicate attractive buying opportunities if companies are approaching peak earnings.”

For now, corporate earnings upgrades are supporting higher stock prices.