Morgan Stanley Wealth Management’s top investment strategist just told the AI bulls something they probably don’t want to hear: the semiconductor trade might be running out of runway.

Lisa Shalett, the firm’s Chief Investment Officer, warned on June 5 that investors are conflating temporary pricing power in chip stocks with genuine, lasting productivity improvements. In a market where everything with “AI” in the pitch deck gets a premium valuation, that distinction matters more than most people think.

The seventh inning stretch

Shalett’s core argument is deceptively simple. The AI capital expenditure boom that has powered semiconductor stocks to eye-watering valuations isn’t in its early stages. It’s closer to what she called the “seventh inning,” borrowing from baseball to suggest we’re much nearer to the end than the beginning.

The numbers backing that thesis are hard to ignore. According to Shalett, the AI-driven rally has accounted for roughly 75% of S&P 500 gains and 80% of its profits over recent periods. Toss in the fact that AI-related spending represents about 90% of recent capital expenditures, and you’ve got a market that’s essentially a one-trick pony wearing a very expensive saddle.

She singled out Broadcom specifically as a warning indicator within the semiconductor space, advising investors to think about trimming exposure to what she views as an overbought sector.

The ghost of Cisco past

Perhaps the most striking element of Shalett’s analysis is her repeated invocation of the “Cisco moment.” For those who weren’t trading during the dot-com era, or who have mercifully blocked it from memory, here’s the translation: Cisco Systems was the darling of the late-1990s internet buildout. Everyone needed routers and networking gear, the stock soared, and then it cratered by more than 80% when the spending cycle ended.

Shalett has been flagging this risk since at least September 2025, when she characterized the post-ChatGPT equity rally as narrow and capex-driven. She warned then that the market faced the potential for a Cisco-style drawdown within 24 months if AI spending momentum faltered. That 24-month clock has been ticking.

Shalett also pointed to declining free cash flow among the dominant cloud providers, the hyperscalers whose AI infrastructure spending has been the primary demand driver for high-end chips.

What this means for crypto and risk assets

Shalett didn’t mention crypto or digital assets in her analysis. But her warnings carry significant implications for anyone holding risk assets, including Bitcoin and the broader crypto market.

Look, nobody’s saying AI is a fad. The technology is transformative. But transformative technologies and good investments aren’t always the same thing at the same time. The internet changed the world, and Cisco shareholders still lost 80% of their money. Shalett’s message isn’t that AI won’t deliver. It’s that the market may have already priced in more delivery than is coming, and the gap between expectation and reality is where portfolios go to get hurt.