The AI supercycle isn’t just holding steady — it’s getting a massive second wind.
In a new note to clients, Morgan Stanley research analyst Shane Brett raised his forecast for semiconductor capital equipment spending to $143 billion in 2026, up from a previous estimate of $136 billion. The new outlook would represent a 23% increase from the prior year.
He also revised his 2027 forecast to $182 billion, up from $161 billion.
This revision paints a picture of an infrastructure race that is accelerating, even as skeptics wonder when these massive capital expenditures will yield a clear return on investment. And according to Brett, the spending increase is creating a new set of winners.
“The market is entering the phase of the cycle where [semiconductor process equipment] returns begin to rival those of memory stocks,” Brett said.
According to Brett, the shift is driven by a massive appetite for two specific types of hardware: memory and logic.
For 2026, he added $5 billion to his forecast for DRAM, or computer memory, and $2 billion for NAND while leaving his estimate for foundry logic unchanged. However, looking toward 2027, he expects a much larger surge, adding $9 billion each to his DRAM and foundry logic forecasts. He suggests these sectors aren’t just producing chips, but are the fundamental building blocks of the data centers powering generative AI.
While the market has long been obsessed with chip designers such as Nvidia (NVDA), Qualcomm (QCOM), Advanced Micro Devices (AMD), and Broadcom (AVGO), Morgan Stanley argues that the real value is now migrating to semiconductor equipment makers — the companies that build the multibillion-dollar machines used in fabrication plants.
Specifically, Brett highlighted Applied Materials (AMAT) as his new “top pick” within the US equipment sector. Despite the company’s crucial role in the supply chain, it is currently trading at a discount compared to rivals like Lam Research (LRCX) and KLA Corp (KLAC). Even with its shares surging roughly 54% year to date — outpacing Lam, up 46%, and KLA, up 28% — the stock has yet to fully reflect its dominant position in the AI memory wave.
There remains concern, though, on how long the current level of AI-driven spending can last. History suggests that if the companies building these machines don’t start showing massive profits soon, the “accelerating” boom Brett describes could eventually fall flat.
And there are broader concerns with demand. On Wednesday, chipmaker Nvidia (NVDA) reported Q4 revenue and a Q1 sales outlook that handily beat Wall Street estimates. Still, the stock tumbled, as investors had high expectations.
