Markets are pricing AI's upside, not its apocalypse

Could artificial intelligence eventually become an existential threat to humanity? It is a question that has resurfaced periodically since the technology's earliest days, and one that recently got Nicholas Colas, co-founder of DataTrek Research, thinking about whether markets should be pricing such a risk today.

His conclusion: probably not.

In a note out Friday, Colas says that if AI ever reaches the point where it threatens civilization, investors are likely to have bigger concerns than where the S&P 500 is trading. More importantly, history suggests markets rarely discount low-probability, potentially catastrophic outcomes before they become tangible.

He points to the Cold War as an example. From the Soviet Union's first nuclear test in 1949 through the early 1980s, the possibility of a devastating nuclear conflict between the United States and the USSR hung over global markets. Yet equity valuations were driven far more by growth, inflation and interest rates than fears of nuclear annihilation.

The Shiller CAPE ratio rose to roughly 24 times in 1965, then a post-Depression high, and fell to around six times by 1982. Colas argues those swings reflected economic conditions, not changing assessments of geopolitical catastrophe.

More broadly, markets tend to react to adverse events once they materialize rather than attempt to price every conceivable worst-case scenario in advance.

That may help explain why investors today appear focused on AI's benefits rather than its risks. The technology is helping drive strong technology-sector earnings, wider profit margins and massive capital spending, all of which are increasingly showing up in economic and corporate results.

In Colas' view, investors are pricing AI's upside, not its downside tail risks. For now, concerns about killer robots can wait.