Investors Are Split on AI Valuations, but Geopolitics Now Ranks as the Bigger Risk

The market argument around artificial-intelligence equities is often framed as bubble versus no bubble. Global Markets Review's October 2026 audience research shows a much less settled picture.

Among 6,530 respondents, 35.9% described AI and semiconductor stocks as overvalued, 34.5% as fairly valued and 29.7% as undervalued. That distribution is too balanced to support a claim that investors have converged on one valuation regime.

Geopolitics outranks valuation as the largest perceived risk

Asked about the greatest risk to AI equities over the next 12 months, geopolitics and export controls were selected more often than valuation. That matters because the semiconductor supply chain is unusually exposed to trade restrictions, manufacturing concentration and cross-border technology policy.

The result also helps explain why a stock can look attractive on earnings growth while still carrying a policy discount that is difficult to model in a conventional price-to-earnings framework.

Analyst targets still matter to most respondents

64.3% said analyst price targets have moderate or major influence on their investment decisions. That does not mean investors follow targets mechanically. It does mean sell-side estimates remain an important input even in a market saturated with direct company disclosures and alternative data.

For GMR, the implication is that analyst coverage should focus less on reproducing targets and more on explaining estimate changes, target dispersion, rating changes and the assumptions that make different targets diverge.

The next useful comparison is valuation versus physical AI investment

AI equity valuations ultimately depend on whether capital spending converts into durable demand and earnings. That means semiconductor analysis increasingly needs to sit beside data-centre power, infrastructure and sovereign investment research.

GMR will use research from its own audience alongside separately sourced reporting from American Commerce Review, Dutch Business Review and Gulf Business Review where those publications have relevant original data on AI infrastructure and investment. Those are editorial citations, not pooled survey samples.

Methodology

The October 2026 findings are based on 6,530 Global Markets Review audience responses. The sample is self-selected and is not representative of all investors.

Future waves will repeat the same valuation, risk and analyst-influence questions to create a longitudinal investor-sentiment series.