Where we heard itCNBC’s Fast Money, Sept. 29, 2026. Speaker: Gene Munster, managing partner at Deepwater Asset Management.
Gene Munster, managing partner at Deepwater Asset Management, says the market is underestimating how much the largest cloud spenders will put into AI next year. He expects a multi-year hardware supercycle running from 2027 to 2029. One of the two public stocks his firm owns has already priced in a lot of that optimism. Micron TechnologyMU closed at $1,063.96 on October 5, 2026, up 466.4% over the past year.
Munster made the comments on CNBC’s Fast Money on September 29, 2026. His figures are his own estimates.

Munster Says Wall Street’s Capex Estimate Is Too Low
His most concrete claim is about spending. By Munster’s account, Wall Street expects hyperscaler capex to grow 40% next year. His own estimate is 60% or better. Both numbers are his own estimates.
Capex forecasts set earnings expectations for everything sold into data centers, including memory. If consensus is too low, the supply chain makes more than investors expect. Micron management agrees. In its latest earnings release, the CEO said “we expect an even stronger fiscal 2027.” On the call, he added: “We do not have line of sight to when supply and demand will return to balance.”

Local Inference Is His 12-Month Theme
Munster pointed to local inference, meaning AI that runs directly on your phone or computer, as the major theme to watch about 12 months out. “That means that the hardware that you have is likely going to have to be upgraded. We’re going to see an AI multi-year supercycle on hardware,” he said. Hardware companies, “as boring as they are,” could have “a great run” over those years, he noted.
Munster made his name covering AppleAAPL, and his argument runs straight through it. On Apple’s July call, then-CEO Tim Cook called rising memory costs “a 100-year flood on the memory pricing.” That same memory is what devices need to run AI locally.
Under 1% Daily Adoption Cuts Both Ways
Munster says less than 1% of Americans use personal AI agents daily, signaling the wave is still early. Low adoption also fits a product people haven’t found a use for yet; the number alone can’t tell which story is true.
A “Go Fast” Message Versus September’s Selloff
Munster said the federal position on AI self-regulation was great news for the AI trade, and he summed up the message as “go fast.” That is his take.
The market reacted differently in mid-September. Global AI stocks fell as industry chiefs called for slowing development, Reuters reported. The Guardian reported that the slide followed calls to slow “reckless” development. Fast Company reported that chip stocks led the selloff. Micron fell 5% after warnings from industry leaders, according to Startup Fortune. Munster says the regulatory mood has switched since then. One analyst called the selloff overblown, citing zero GPU slowdown, according to finance.yahoo.com.
Considering a Bold Forecast Against a Huge Run
Micron is up 272.8% this year. Apple closed at $332.89 on October 5, up 29.0% over the past year and 22.4% this year. Munster’s core claim is that spending is underestimated, yet the clearest stock linked to that view has already had an extraordinary run. He could still be right if earnings keep up, but the argument and valuation background are very different today than a year ago.
Micron’s fiscal fourth quarter at a glance
Munster disclosed that his firm owns both Apple and Micron. Anyone publicly forecasting a hardware supercycle while holding a memory stock has an interest in that view being correct.
A panelist cited economist Robert Solow’s 1987 observation that the computer age was visible everywhere except in productivity statistics, and noted that earlier general-purpose technologies took a long time to reach broad adoption.
Two Calls Investors Can Check
Munster’s capex call has a defined window, and hyperscaler guidance will settle it. His local inference call has a stated 12-month horizon. Both can be checked against what actually happens.

Gene Munster sees Wall Street badly underestimating what the biggest tech spenders will pour into AI next year, and he owns the memory stock that sits at the center of his bet. Whether the case still holds after a historic run is the question every investor needs to answer.
Munster thinks Wall Street is lowballing next year’s AI spending and expects AI that runs on phones and computers to drive a hardware upgrade cycle from 2027 to 2029. His firm owns Micron, a direct beneficiary. The risk is the price: Micron rose about 466% in the year before this story, so much of that good news may already be counted.